Hybrid Artificial Intelligence and IoT in Healthcare (Intelligent Systems Reference Library, 209) 9811629714, 9789811629716

Citation preview

Intelligent Systems Reference Library 209

Akash Kumar Bhoi Pradeep Kumar Mallick Mihir Narayana Mohanty Victor Hugo C. de Albuquerque   Editors

Hybrid Artificial Intelligence and IoT in Healthcare

Intelligent Systems Reference Library Volume 209

Series Editors Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland Lakhmi C. Jain, KES International, Shoreham-by-Sea, UK

The aim of this series is to publish a Reference Library, including novel advances and developments in all aspects of Intelligent Systems in an easily accessible and well structured form. The series includes reference works, handbooks, compendia, textbooks, well-structured monographs, dictionaries, and encyclopedias. It contains well integrated knowledge and current information in the field of Intelligent Systems. The series covers the theory, applications, and design methods of Intelligent Systems. Virtually all disciplines such as engineering, computer science, avionics, business, e-commerce, environment, healthcare, physics and life science are included. The list of topics spans all the areas of modern intelligent systems such as: Ambient intelligence, Computational intelligence, Social intelligence, Computational neuroscience, Artificial life, Virtual society, Cognitive systems, DNA and immunity-based systems, e-Learning and teaching, Human-centred computing and Machine ethics, Intelligent control, Intelligent data analysis, Knowledge-based paradigms, Knowledge management, Intelligent agents, Intelligent decision making, Intelligent network security, Interactive entertainment, Learning paradigms, Recommender systems, Robotics and Mechatronics including human-machine teaming, Self-organizing and adaptive systems, Soft computing including Neural systems, Fuzzy systems, Evolutionary computing and the Fusion of these paradigms, Perception and Vision, Web intelligence and Multimedia. Indexed by SCOPUS, DBLP, zbMATH, SCImago. All books published in the series are submitted for consideration in Web of Science.

More information about this series at http://www.springer.com/series/8578

Akash Kumar Bhoi · Pradeep Kumar Mallick · Mihir Narayana Mohanty · Victor Hugo C. de Albuquerque Editors

Hybrid Artificial Intelligence and IoT in Healthcare

Editors Akash Kumar Bhoi Department of Computer Science and Engineering Sikkim Manipal Institute of Technology (SMIT), Sikkim Manipal University Majitar, Sikkim, India Mihir Narayana Mohanty Department of Electronics and Communication Engineering Siksha ‘O’ Anusandhan (Deemed to be University) Bhubaneswar, Odisha, India

Pradeep Kumar Mallick School of Computer Engineering Kalinga Institute of Industrial Technology Deemed to be University Bhubaneswar, Odisha, India Victor Hugo C. de Albuquerque Graduate Program on Teleinformatics Engineering Federal University of Ceará Fortaleza, Brazil

ISSN 1868-4394 ISSN 1868-4408 (electronic) Intelligent Systems Reference Library ISBN 978-981-16-2971-6 ISBN 978-981-16-2972-3 (eBook) https://doi.org/10.1007/978-981-16-2972-3 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

Hybrid artificial intelligence (AI) with bio-inspired techniques, genetic algorithm, neuro-fuzzy algorithms and soft computing approaches significantly improves the prediction of critical cardiovascular abnormalities and other healthcare solutions to the ongoing challenging research. AI utilizes measurable methods to enable computer frameworks to “learn” with approaching information and to distinguish examples and settle on choices with negligible human bearing. The extent of conversation of the potential ramifications of AI and IoT in future human services is practically boundless. This book intended to cover the applications for hybrid AI and IoT for an integrated approach and problem solving in the areas of radiology, drug interactions, creation of new drugs, imaging, electronic health records, disease diagnosis, telehealth and mobility-related problems in healthcare. Chapter “Hybrid Cloud/Fog Environment for Healthcare: An Exploratory Study, Opportunities, Challenges, and Future Prospects” discusses the areas of applicability in healthcare systems of hybrid cloud/fog computing. The several extraordinary opportunities brought by the technologies in the healthcare system with research challenges in deployment are discussed. Chapter “Hybrid Intelligent System for Medical Diagnosis in Health Care” by Moolchand Sharma et al. focuses on the need for a hybrid intelligent system in the healthcare industry and their medical diagnosis applicability. Chapter “Remote Patient Monitoring Using IoT, Cloud Computing and AI” illustrates the concept of cloud and AI-based IoT for remote healthcaring. It also discusses different decision-making systems using AI and the principle of operation of several cloud infrastructures used to access secured medical records. Chapter “An Analytical Study of the Role of M-IoT in Healthcare Domain” by Bidisha Chanda et al. brings to light several applications of M-IoT in healthcare in measuring body temperature, monitoring blood glucose level, ECG, etc. The effect of Internet of things has been evolving in every aspect of living; however, the effect it has at the healthcare industry is massive because of its growing need and accuracy. Chapter “Hybrid AI and IoT Approaches Used in Health Care for Patients Diagnosis” proposes the AIWAC emotion interaction system, which has been designed to build

v

vi

Preface

the cognitive low-power wide area zone (LPWAN) and further test the proposed AIenabled LPWA hybrid method. Chapter “RADIoT: The Unifying Framework for IoT, Radiomics and Deep Learning Modeling” by Sakinat Oluwabukonla FOLORUNSO et al. discusses the different types and sources of radiological data, feature extraction and selection method for image analysis. The chapter also presents different ML models ideal for the radiomics and parameter tuning. Chapter “Hybrid Artificial Intelligence and IoT in Health care for Cardiovascular Patient in Decision-Making System” discusses on how to create an IoT device that collects body area sensor (BAS) data with the intent of providing early warning about an imminent heart arrest. C. Aravindan et al. in the chapter “A Smart Assistive System for Visually Impaired to Inform Acquaintance Using Image Processing (ML) Supported by IoT” proposed to provide a new interface support system for a visually impaired person to independently handle everyday life activities like recognizing and greeting a friend during a walk in the street. Chapter “Internet of Things in Health Care: A Survey” by Ahmed Izzat Alsalibi et al. surveys current advances in IoT-based healthcare focused in the aspects of privacy and security techniques, e-health and m-health, system design and architecture, cloud, fog and evolutionary computing, and network and communication techniques. Chapter “Disease Diagnosis System for IoT-Based Wearable Body Sensors with Machine Learning Algorithm” proposes a framework for IoT-WBN based with a machine learning algorithm (ML). The data collected from different wearable sensors like body temperature, glucose sensors, heartbeat sensors and chest were transmitted through IoT devices to the integrated cloud database. Chapter “Integration of Machine Learning and IoT in Healthcare Domain” discusses on the integration of machine learning and IoT in healthcare domain. The combination of these two technologies can be of great help in clinical sector in the generation of a receptive and inter-related environment, thereby providing various services to healthcare staffs and patients. Adriana Trus, culescu et al. in the chapter “Managing Interstitial Lung Diseases with Computer-Aided Visualization” presented flexible modeling of diffuse lung interstitial diseases (DLID), which is especially important in Idiopathic pulmonary fibrosis (IPF), in which the accepted computer tomography diagnosis criteria allow extremely diverse individual variations. Chapter “Use of Machine Learning Algorithms to Identify Sleep Phases Starting from ECG Signals” by Giovanna Sannino and Ivanoe De Falco presents the identification of the different sleep phases a subject is experiencing by using heart rate variability (HRV) values. These are computed starting from the signals gathered from electrocardiogram (ECG) sensors placed on the subject and online classification in an IoT-based fully automated e-health system. Chapter “Emerging Technologies for Pandemic and Its Impact” discusses about the emerging technologies for pandemic and its impact and their remedies with the latest technological developments whether it is the security, tracking, IoT-based healthcare systems, healthcare seminars, conferences, etc. Finally, the chapter “Impact of Artificial Intelligence in Health care: A Study” presents the role of AI in this critical healthcare sector along with some popular existing research works in the healthcare domain. Software projects involving AI in

Preface

vii

this sector are also summarized, and a real-time implementation of medical imaging using different computational methods is demonstrated. The editors wish to thank all the authors who have contributed to this book and sincerely thank the Springer Nature editorial and production team for the constant support. We send our best wishes to our readers! Majitar, India Bhubaneswar, India Bhubaneswar, India Fortaleza, Brazil

Akash Kumar Bhoi Pradeep Kumar Mallick Mihir Narayana Mohanty Victor Hugo C. de Albuquerque

Contents

Hybrid Cloud/Fog Environment for Healthcare: An Exploratory Study, Opportunities, Challenges, and Future Prospects . . . . . . . . . . . . . . . Joseph Bamidele Awotunde, Akash Kumar Bhoi, and Paolo Barsocchi

1

Hybrid Intelligent System for Medical Diagnosis in Health Care . . . . . . . Moolchand Sharma, Akanksha Kochhar, Deepak Gupta, and Jafar Al Zubi

21

Remote Patient Monitoring Using IoT, Cloud Computing and AI . . . . . . M. V. V. Prasad Kantipudi, C. John Moses, Rajanikanth Aluvalu, and Sandeep Kumar

51

An Analytical Study of the Role of M-IoT in Healthcare Domain . . . . . . . Bidisha Chanda, Pradeep Kumar Mallick, and Gyoo-Soo Chae

75

Hybrid AI and IoT Approaches Used in Health Care for Patients Diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shruti Mathur, Akhilesh Kumar Sharma, and Phayung Meesad

97

RADIoT: The Unifying Framework for IoT, Radiomics and Deep Learning Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Sakinat Oluwabukonla Folorunso, Joseph Bamidele Awotunde, Femi Emmanuel Ayo, and Khadijah-Khuburah Adebisi Abdullah Hybrid Artificial Intelligence and IoT in Health care for Cardiovascular Patient in Decision-Making System . . . . . . . . . . . . . . . . 129 M. Safa, A. Pandian, T. Kartick, K. Chakrapani, G. Geetha, and G. Saranya A Smart Assistive System for Visually Impaired to Inform Acquaintance Using Image Processing (ML) Supported by IoT . . . . . . . . 149 C. Aravindan, R. Arthi, R. Kishankumar, V. Gokul, and S. Giridaran Internet of Things in Health Care: A Survey . . . . . . . . . . . . . . . . . . . . . . . . . 165 Ahmed Izzat Alsalibi, Mohd Khaled Yousef Shambour, Muhannad A. Abu-Hashem, Mohammad Shehab, and Qusai Shambour

ix

x

Contents

Disease Diagnosis System for IoT-Based Wearable Body Sensors with Machine Learning Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 Joseph Bamidele Awotunde, Sakinat Oluwabukonla Folorunso, Akash Kumar Bhoi, Paul Olujide Adebayo, and Muhammad Fazal Ijaz Integration of Machine Learning and IoT in Healthcare Domain . . . . . . . 223 Ananya Chattopadhyay, Sushruta Mishra, and Alfonso González-Briones Managing Interstitial Lung Diseases with Computer-Aided Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 Adriana Trus, culescu, Laura Broasc˘a, Versavia Maria Ancus, a, Diana Manolescu, Emanuela Tudorache, and Cristian Oancea Use of Machine Learning Algorithms to Identify Sleep Phases Starting from ECG Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 Giovanna Sannino and Ivanoe De Falco Emerging Technologies for Pandemic and Its Impact . . . . . . . . . . . . . . . . . 291 Dinesh Kumar Saini, Sandhiya Bala, Akhilesh Kumar Sharma, and Kashif Zia Impact of Artificial Intelligence in Health care: A Study . . . . . . . . . . . . . . . 311 Devyani Bairagya, Hrudaya Kumar Tripathy, Akash Kumar Bhoi, and Paolo Barsocchi

Editors and Contributors

About the Editors Akash Kumar Bhoi (B.Tech., M.Tech., Ph.D.) is working as Assistant Professor (Research) in the Department of Computer Science and Engineering at Sikkim Manipal Institute of Technology (SMIT), India, since 2012. He is also working (Period: 20 January 2021– 19 January 2022) as Research Associate at Wireless Networks (WN) Research Laboratory, Institute of Information Science and Technologies, National Research Council (ISTI-CRN) Pisa, Italy. He is a University Ph.D. Course Coordinator for “Research & Publication Ethics (RPE).” He is a member of IEEE, ISEIS, and IAENG, an associate member of IEI, UACEE, and an editorial board member reviewer of Indian and international journals. He is also a regular reviewer of repute journals, namely IEEE, Springer, Elsevier, Taylor and Francis, Inderscience, etc. His research areas are Biomedical Technologies, the Internet of Things, Computational Intelligence, Antenna, Renewable Energy. He has published several papers in national and international journals and conferences. He has 100+ documents registered in the Scopus database by the year 2020. He has also served on numerous organizing panels for international conferences and workshops. He is currently editing several books with Springer Nature, Elsevier and Routledge and CRC Press. He is also serving as Guest editor for special issues of the journal like Springer Nature and Inderscience.

xi

xii

Editors and Contributors

Dr. Pradeep Kumar Mallick is currently working as Senior Associate Professor in the School of Computer Engineering, Kalinga Institute of Industrial Technology (KIIT) Deemed to be University, Odisha, India .He has also served as Professor and Head Department of Computer Science and Engineering , Vignana Bharathi Institute of Technology, Hyderabad . He has completed his Post Doctoral Fellow (PDF) from Kongju National University South Korea , Ph.D. from Siksha ‘Ó’ Anusandhan University, M.Tech. (CSE) from Biju Patnaik University of Technology (BPUT), and MCA from Fakir Mohan University Balasore, India. Besides academics, he is also involved in various administrative activities, Member of Board of Studies to C. V. Ramman Global University Bhubaneswar, Member of Doctoral Research Evaluation Committee, Admission Committee etc. His area of research includes Data Mining, Image Processing, Soft Computing, and Machine Learning. Now he is the editorial member of Korean Convergence Society for SMB .He has published 12 edited books, one text Book and more than 80 research papers in National and international journals and conference proceedings to his credit. Dr. Mihir Narayana Mohanty is presently working as Professor in the Department of Electronics and Communication Engineering, Institute of Technical Education and Research, Siksha ‘O’ Anusandhan University, Bhubaneswar, Odisha. He has published over 300 papers in International/National Journals, Book Chapters, and Conferences along with approximately 25 years of teaching experience in UG and PG level. He is the active member of many professional societies like IEEE, IET, ISTE, IRED, EMC and EMI Engineers India, ISCA, ACEEE, IAEng, etc. Also he is a Fellow of IE (I) and IETE. He has received his M.Tech. Degree in Communication System Engineering from Sambalpur University, Sambalpur, Odisha. Also he has done his Ph.D. work in Applied Signal Processing. He was working as Associate Professor and Head in the Department of Electronics and Instrumentation Engineering, Institute of Technical Education and Research, Siksha ‘O’ Anusandhan Deemed to be University, Bhubaneswar, Odisha. His area of research interests includes—Applied Signal and Image Processing,

Editors and Contributors

xiii

Digital Signal/Image Processing, Biomedical Signal Processing, Microwave Communication Engineering and Bioinformatics. He has worked as Guest lecturer in many universities. Simultaneously has given many invited talks. He has reviewed many Springer and IEEE based conference papers as well as for some International Journal Papers. Currently seven research scholars along with two number of PG students are under his guidance on Antenna, Signal, Image & Speech Processing and Communication Engineering. Victor Hugo C. de Albuquerque (M’17, SM’19) is a collaborator Professor and senior researcher at the Graduate Program on Teleinformatics Engineering at the Federal University of Ceará, Brazil. He has a Ph.D. in Mechanical Engineering from the Federal University of Paraíba (UFPB, 2010), an M.Sc. in Teleinformatics Engineering from the Federal University of Ceará (UFC, 2007), and he graduated in Mechatronics Engineering at the Federal Center of Technological Education of Ceará (CEFETCE, 2006). He is a specialist, mainly, in Image Data Science, IoT, Machine/Deep Learning, Pattern Recognition, Robotic.

Contributors Khadijah-Khuburah Adebisi Abdullah Department of Mathematical Science, Olabisi Onabanjo University, Ago-Iwoye, Nigeria Muhannad A. Abu-Hashem Department of Geomatics, Faculty of Architecture and Planning, King Abdulaziz University KAU, Jeddah, Saudi Arabia Paul Olujide Adebayo Department of Computer Science, Federal Polytechnic, Nasarawa, Nigeria Ahmed Izzat Alsalibi Department of Information Technology, Faculty of Engineering and Information Technology, Israa University, Gaza, Palestine Rajanikanth Aluvalu Vardhaman College of Engineering, Hyderabad, India Versavia Maria Ancus, a Department of Computer and Information Technology— Automation and Computers Faculty, “Politehnica” University of Timisoara, Timis, oara, RO, Romania C. Aravindan Department of ECE, SRM Institute of Science and Technology, Chennai, Tamil Nadu, India

xiv

Editors and Contributors

R. Arthi Department of ECE, SRM Institute of Science and Technology, Chennai, Tamil Nadu, India Joseph Bamidele Awotunde Department of Computer Science, University of Ilorin, Ilorin, Nigeria Femi Emmanuel Ayo Department of Computer Science, McPherson University, Ogun State, Nigeria Devyani Bairagya School of Computer Engineering, Kalinga Institute of Industrial Technology (KIIT), Deemed To Be University, Bhubaneswar, Odisha, India Sandhiya Bala Nielsen Global Connect Chennai, Chennai, India Paolo Barsocchi Institute of Information Science and Technologies, National Research Council, Pisa, Italy Akash Kumar Bhoi Department of Computer Science and Engineering, Sikkim Manipal Institute of Technology (SMIT), Sikkim Manipal University (SMU), Majitar, Sikkim, India; Institute of Information Science and Technologies, National Research Council, Pisa, Italy Laura Broasc˘a Department of Computer and Information Technology—Automation and Computers Faculty, “Politehnica” University of Timisoara, Timis, oara, RO, Romania Gyoo-Soo Chae Division of Smart IT Engineering, Baekseok University, Cheonansi, South Korea K. Chakrapani Department of Information Technology, School of Computing, SASTRA Deemed University, Tanjore, Tamil Nadu, India Bidisha Chanda School of Computer Engineering, Kalinga Institute of Industrial Technology, Deemed to be University, Bhubaneswar, India Ananya Chattopadhyay School of Computer Engineering, KIIT Deemed To Be University, Bhubaneswar, India Ivanoe De Falco Institute for High-Performance Computing and Networking (ICAR), National Research Council of Italy (CNR), Naples, Italy Sakinat Oluwabukonla Folorunso Department of Mathematical Science, Olabisi Onabanjo University, Ago-Iwoye, Nigeria G. Geetha School of Computing, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India S. Giridaran Department of ECE, SRM Institute of Science and Technology, Chennai, Tamil Nadu, India V. Gokul Department of ECE, SRM Institute of Science and Technology, Chennai, Tamil Nadu, India

Editors and Contributors

xv

Alfonso González-Briones Research Group on Agent-Based, Social and Interdisciplinary Applications (GRASIA), Complutense University of Madrid, Madrid, Spain; BISITE Research Group, University of Salamanca, Salamanca, Spain; Air Institute, IoT Digital Innovation Hub, Salamanca, Spain Deepak Gupta Maharaja Agrasen Institute of Technology, MAIT, Delhi (GGSIPU), India Muhammad Fazal Ijaz Department of Intelligent Mechatronics Engineering, Sejong University, Seoul, South Korea M. V. V. Prasad Kantipudi Sreyas Institute of Engineering and Technology, Hyderabad, India T. Kartick School of Computing, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India R. Kishankumar Department of ECE, SRM Institute of Science and Technology, Chennai, Tamil Nadu, India Akanksha Kochhar Maharaja Agrasen Institute of Technology, MAIT, Delhi (GGSIPU), India Sandeep Kumar Sreyas Institute of Engineering and Technology, Hyderabad, India Pradeep Kumar Mallick School of Computer Engineering, Kalinga Institute of Industrial Technology, Deemed to be University, Bhubaneswar, India; Division of Smart IT Engineering, Baekseok University, Cheonan-si, South Korea Diana Manolescu Department of Radiology, Center for Research and Innovation in Precision Medicine of Respiratory Diseases (CRIPMRD), “Victor Babes, ” University of Medicine and Pharmacy, Timis, oara, RO, Romania Shruti Mathur Manipal University Jaipur, Jaipur, Rajasthan, India Phayung Meesad King Mongkuts University Thailand NB, Bangkok, Thailand Sushruta Mishra School of Computer Engineering, KIIT Deemed To Be University, Bhubaneswar, India C. John Moses Sreyas Institute of Engineering and Technology, Hyderabad, India Cristian Oancea Department of Pneumology, Center for Research and Innovation in Precision Medicine of Respiratory Diseases (CRIPMRD), “Victor Babes, ” University of Medicine and Pharmacy, Timis, oara, RO, Romania A. Pandian School of Computing, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India M. Safa School of Computing, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India Dinesh Kumar Saini Manipal University Jaipur, Jaipur, India

xvi

Editors and Contributors

Giovanna Sannino Institute for High-Performance Computing and Networking (ICAR), National Research Council of Italy (CNR), Naples, Italy G. Saranya School of Computing, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India Mohd Khaled Yousef Shambour The Custodian of the Two Holy Mosques Institute for Hajj and Umrah Research, Umm Al-Qura University, Mecca, Saudi Arabia Qusai Shambour Department of Software Engineering, Faculty of InformationTechnology, Al-Ahliyya Amman University, Amman, Jordan Akhilesh Kumar Sharma Manipal University Jaipur, Jaipur, Rajasthan, India Moolchand Sharma Maharaja Agrasen Institute of Technology, MAIT, Delhi (GGSIPU), India Mohammad Shehab Department of Software Engineering, The World Islamic Science and Education University, Amman, Jordan Hrudaya Kumar Tripathy School of Computer Engineering, Kalinga Institute of Industrial Technology (KIIT), Deemed To Be University, Bhubaneswar, Odisha, India Adriana Trus, culescu Department of Pneumology, Center for Research and Innovation in Precision Medicine of Respiratory Diseases (CRIPMRD), “Victor Babes, ” University of Medicine and Pharmacy, Timis, oara, RO, Romania Emanuela Tudorache Department of Pneumology, Center for Research and Innovation in Precision Medicine of Respiratory Diseases (CRIPMRD), “Victor Babes, ” University of Medicine and Pharmacy, Timis, oara, RO, Romania Kashif Zia Sohar University Oman, Sohar, Oman Jafar Al Zubi School of Engineering, Al-Balqa Applied University, Jordan, India

Hybrid Cloud/Fog Environment for Healthcare: An Exploratory Study, Opportunities, Challenges, and Future Prospects Joseph Bamidele Awotunde , Akash Kumar Bhoi , and Paolo Barsocchi

Abstract The healthcare system has been on the frontline in recent years, and new technologies have greatly benefited healthcare. Researchers have tried to find solutions to different problems associated with the healthcare system by applied various modern technologies approaches. Among the various technologies, are fog and computing used in smart healthcare systems. These applications with the Internet of things (IoT) recently have help in dispersed patient data globally and have advanced healthcare systems. Hence, various applications and solutions using cloud computing have been proposed by researchers to manage healthcare statistics. However, the issues of latency, context-awareness, and a huge volume of data are remaining challenges in cloud computing. Hence, the possibility of transmission errors and the probability of delay in data processing remain a problem as healthcare datasets become more complex and larger. The most alternative solution to those challenges is fog computing in reducing data management complexity in the healthcare system, thus increasing reliability. But, before using fog computing, it is very essential to look into its associated challenges in other to manage healthcare data effectively. Therefore, this chapter discusses the areas of applicability in healthcare systems of hybrid cloud/fog computing. The several extraordinary opportunities brought by the technologies in the healthcare system with research challenges in deployment are discussed. The applications of fog in IoT-based devices bring healthcare components in a distant cloud operating nearer to data sources and the end-users, thus, resulting in context-awareness and lower latency.

J. B. Awotunde (B) Department of Computer Science, University of Ilorin, Ilorin, Nigeria e-mail: [email protected] A. K. Bhoi Department of Computer Science and Engineering, Sikkim Manipal Institute of Technology (SMIT), Sikkim Manipal University (SMU), Majitar, Sikkim 737136, India A. K. Bhoi · P. Barsocchi Institute of Information Science and Technologies, National Research Council, 56124 Pisa, Italy e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 A. Kumar Bhoi et al. (eds.), Hybrid Artificial Intelligence and IoT in Healthcare, Intelligent Systems Reference Library 209, https://doi.org/10.1007/978-981-16-2972-3_1

1

2

J. B. Awotunde et al.

Keywords Hybrid cloud · Fog computing · Healthcare system · Internet of things · Latency · Technologies

1 Introduction The healthcare sector faces challenges including budget pressures, management of various service delivery, and growth in aging populations [1]. The increased use of information and communications technology (ICT) will help the healthcare sector reduce if not overcome these challenges [2]. In connection to the need to make service delivery more effective, ICT innovations have contributed to higher healthcare technology adoption [3, 4]. To help healthcare, ICT has been used to provide greater access to medical records [5] and the area of decision making in the healthcare system; ICT has additional ability to help minimize costs and enhance service deliveries [2, 5, 6]. One of the world’s most recent groundbreaking technology is cloud computing (CC), and the applications are increasingly growing daily. The applications CC is widely used in smart healthcare systems having storage devices capacity to stored big data generated by the IoT-based devices daily, a significant part of the services have been shifted to the cloud. This has help healthcare systems to provide costeffective and reliable services globally. The CC has brought an initiative to migrate a widespread perception that such cloud boundaries and security problems will obstruct the transition. To provide better services with an avoidable price to patients, the healthcare system has been a move to the cloud globally with better healthcare facilities. The need for a greater low cost with effective healthcare system necessitates the need for cloud computing in the medical system, thus explain why healthcare providers’ services switching to the cloud. According to the report, various branched of medical sciences have benefited from this paradigm. For example, cardiology has benefited from efficient data storage and retrieval systems in accessing historical and current patient data on the cloud database. The maintenance of CC software licensing that is very low about other medical software contributed to the increased popularity. In the delivering of an efficient information technology (IT), CC offers an innovative system and play paramount roles. The technology is identified as one of the major technology to enhance healthcare service deliveries using IoT-based devices in inpatient data collections. When combines IoT and CC technologies, they benefited in an equal manner with mutual understanding in handling medical information. Developed a monitoring system combining Cloud and Edge computing technologies has been proved efficient in the areas of providing healthcare facilities in remote areas, thus helps the caregivers and physicians to provide qualities healthcare services to their patients. The CC is used as a supportive technology in an IoT-based system in terms of computational capability, storage, resource utilization, and reduced energy consumption. Also, the cloud has been a favorite of IoT-based system by enhancing service deliveries globally and deliver unspeakable services in a distributed and

Hybrid Cloud/Fog Environment for Healthcare … Table 1 Appraisal of cloud and fog computing

Requirements

3 Cloud computing

Fog computing

Mark User

Internet Users

Mobile users

Location of servers

Within Internet

Edge nodes

Service type

Global information

Localized information services

Geographical distribution

Centralized

Distributed

Distance between client and server

Multiple hops

Single hop

Delay jitter

High

Low

Latency

High

Low

Type of connectivity Leased line

Wireless

Location awareness

No

Yes

Server nodes

Few

Large

N/W bandwidth

More

Low

Reply time

Minutes

Milliseconds, sub-seconds

Security

Less secure

Very secure

dynamic manner. The IoT-based cloud framework can still be extended in the smart environment for the development and application of new service delivery. Table 1 compared the cloud and fog computing using IoT-based requests. The huge amount of big data generated by IoT-based devices can no longer be handle by the CC again, new powerful computing models are required. The security concerns, low latency, speedy processing requirements need new powerful computing techniques to best place processing, conserve network bandwidth, and making IoTbased system operates in a reliable environment [7]. All these IoT-based system requirements can never be met with the traditional CC architectures alone; therefore, a better and powerful computing model is required. Latency creates dominant strategy by transfers data for processing from the network edge to the data center. Bandwidth is quickly outpaced by traffic from thousands of users. Also, the cloud servers neglect other protocols the IoT devices use and interact only with IP. The best location for most IoT data to be analyzed is close to the machines that generate and function on that data, and this is called computing with fog. To close the gap and bridge linking IoT-based devices, a powerful technology called fog models has been employed to help the IoT-based devices to process the big data generated from the patient. The processes at the edge of the computational resources become easy with the fog model to handle and outline the data from IoT-based devices, thus, become easy and greatly improved. The idea of cloud computing is very similar to fog computing since both are built with virtual systems and offer many of the similar architectures and features that facilitate the versatility and scalability of computing, storage, and networking resources on-demand

4

J. B. Awotunde et al.

supplies. Although with the emerging trend in networking in terms of demand, the two technologies have a wide barrier. The businesses and the end-users are free to use cloud computing from defining certain specifics, such as storage capacity, limits on computing, and the cost of network connectivity. The problem of latency sensitive applications in real-time still arises for nodes to meet their delay requirements is still arising [8, 9]. The issue of security of these huge volumes of data should also be the main concern for any business-minded experts because the problem causes to their reputation and they are constrained by the law to keep all data safe. The cloud-based provides the liberty of accessing data from the service providers anytime in any part of the world, hence, exposure the IoT-based data to security and privacy threats. In medical practice, the use of cloud/fog computing has increased tremendously. To gain more accurate diagnosis results, cloud/fog has been widely used to decrease the burden on the medical experts and help in decreasing the decision time of traditional methods of the diagnosis process. There are significant and improvements in the treatment, prediction, screening, drug/vaccine development processes, and application of medication in healthcare sectors with continuing expansion in cloud/fog computing. Their applications have reduced human intervention in medical processes, and the cost of medical applications has reduced. Fog layer has moved the cloud closer to end-users, and the data center can now be process close to the network nodes. The computational at the edge of network nodes becomes easier and the cloud storage has greatly expanded by the fog models. This triggers the idea of a cloud-focused cutoff and embarks on how the sequence of data generated by IoT-based devices can be stored and operated. CISCO introduced another version of fog computing at the edge of the network that can used billions of connected devices application strenuously [10, 11]. Fogging as it is called is a distributed database that allow a remote data center manages its services at the network edge, and the program runs on them. Fog is about the real world being dealt with. The edge or fog paradigm addresses the problems with the basic concept of finding small servers in the vicinity of users and devices called edge servers [12]. Therefore, this chapter presents both applications of CC and fog in smart healthcare system; also, their applicability and challenges were also discussed. The deployment of cloud and fog computing has greatly helped the smart healthcare system.

2 Applications of Cloud Computing in Smart Healthcare System Cloud computing will play an essential part in absorbing healthcare transformation expenses, optimizing assets, and bringing the new age of technology to life. Emerging policies are targeted at obtaining data at anytime, anywhere that can be achieved by transferring health data to the cloud. This contemporary distribution model will make healthcare more productive and operational and lowering the price of innovation

Hybrid Cloud/Fog Environment for Healthcare …

5

expenditures [13], but it also presents some obstacles due to issues regarding the security of sensitive health information and compliance with a specific criterion such as HIPAA. Healthcare providers, taking into account these security and privacy concerns, can unquestionably reap the benefits of cloud computing technologies and provide substantial benefits, such as assisting to expand the eminence of service for patients and reducing healthcare spending [14]. Cloud computing’s critical features are (1) self-service on-demand, (2) huge amount of database, (3) allow devices sharing, (4) swift elasticity, and (5) calculated facilities. In complex resources, clouds offer benefits such as processing energy or storage abilities, universal access to resources from anywhere at any time, and high resource versatility and scalability. In several business fields, these advantages have been the purpose for the growing acceptance of cloud computing. This principle has also evidently been adopted in the area of healthcare in recent years. The cloud with IoT-based online application works efficiently over the ordinary cloud-based applications. The medical, bank, and military are instance of sectors that have benefited greatly from emerging models. The cloud-based IoT-based system has helped in providing medical sectors with effective services in the areas of monitoring and access of medical records remotely. IoT-centric application has been used to collect real time from patients using devices and sensors. Also, artificial intelligence can be used to analyzed data capture using IoT-based devices for effective diagnosis of disease at the correct time before the illness reach advanced stages. The CC has the capacity to share data between different systems within IoT-based system. This capacity is something that IT urgently needs for healthcare. Cloud computing, for example, can enable healthcare professionals to share data such as EHR, doctor’s references, medications, insurance data, research reports stored via various information systems. In the radiological market, where many organizations have switched to the cloud to minimize their computing costs and promote the sharing of pictures, this is already happening [15]. Cloud computing has provided clinics, hospitals, insurance providers, pharmacies, and other healthcare companies the ability to agree to cooperate and exchange healthcare data to provide improved service quality and minimize costs. Looking at the developments in the industry, it seems that once all the obstacles it presents are resolved, cloud-based schemes will eventually turn out to be the standard in healthcare. The healthcare system’s ecosystem, which comprises health insurance providers, hospital and physician networks, laboratories, clinics, patients, and other institutions, is vast, diverse and highly nuanced [16]. And all of these must operate under many government regulations [17]. To function efficiently and rapidly in this ecosystem, any sensitive details must be exchanged confidentially and in a safe way between these agencies quickly and accurately. In the healthcare sector, protecting the patient’s data is known to be very sensitive to privacy issues. Possibly, one of the reasons why the development of healthcare moving into the cloud has negatively affected. Innovative technology and resources must be managed when it comes to cloud sharing. However, as they theoretically range between cities, states, and even nations, many other records, knowledge, and resources can benefit from collaboration by cloud usage.

6

J. B. Awotunde et al.

Private clouds tend to be deployed first because of security issues in the current scenario and then shift into public networks [16]. It may be a good idea to first set out the healthcare industry’s top priorities and then analyze which elements of cloud computing can be efficiently implemented to support them. The efficiency of services provided to patients and customers, privacy, data security and integrity, and catastrophe recovery seems to be at the forefront of today’s rising global healthcare costs [17, 18]. Some of the inherent features can be leveraged to meet some of these objectives, such as flexible architecture, data centers for the provision of permanent data, protection models, and rapid access to information, among others. Cloud computing encourages IT facilities that are accessible from all locations and at all times [19]. It is a new mechanism, not a new technology, to deliver computing services [20]. Examples of non-medical cloud services are Microsoft Office 365 and Google Docs, while examples of medical service apps are Microsoft HealthVault and Google Health platforms [21]. Compared to traditional computing, there are three major enhancements provided by the cloud computing model: (1) computationally intensive solutions are accessible on request, (2) service delivery without charge. Customer upfront commitment requirements, and (3) flexibility for short-term use [22]. Many sectors have been influenced by the cloud model, and it is estimated that approximately, 80% of today’s businesses will have embraced cloud computing by 2020 [15]. Besides, companies that lack capital and infrastructure should implement cloud computing to set up on-site applications [23]. Cloud computing, especially within the electronic health records (EHRs) field, is transforming healthcare IT [24]. Cost minimization in IT investments will contribute to improved healthcare facilities [25] and estimated that drug costs can be decreased by 80% and payment can be done within 2 h for patients and insurance providers as compared to up to seven days with an implementation. To dynamically compute patient records with sensors that are attached to medical equipment to process data for collection, accessibility, and distribution, a cloud-based framework has been suggested. During any disease outbreak, this device can reduce typical errors or data collection errors manually [26, 27], not just simplification of the procedure, but also increased access to high-quality data [28]. By combining the ambulance services with patient records, the Greek National Health Service has built an emergency care program in the cloud, ensuring direct access for doctors while being willing to use all resources while maintaining low costs as much as possible [29]. In Australia, as a partnership between Telstra and the Royal Australian College of General Practitioners (RACGP) suggested an e-health cloud [15]. The goal of this collaboration is to develop diagnostic and response to situations, medical software, medications, and training and referral facilities. Cloud processing technologies have provided successful support for bioinformatics research in the medical field [15, 30, 31]. Although cloud computing has several value-added ideas driven by a novel paradigm of IT service distribution over the network, economic benefits appear to be the most significant factor in its popularity and widespread acceptance. Lowering the cost of healthcare delivery is a significant catalyst for the implementation of cloud technology in healthcare. This expense has risen to such immense

Hybrid Cloud/Fog Environment for Healthcare …

7

proportions that governments are facing severe problems with funding. The realization that patient care can be enhanced by technology while lowering costs has ensured that policymakers can drive the historically sluggish healthcare sector to a faster rate of adoption. Big data development in healthcare is another significant factor [32]. When the quantity of digital information grows, the capacity to manage this information is becoming an increasing challenge. This knowledge embraces the keys to prospective clinical developments but is also inaccessible to scientists. Cloud computing can be the supporting reason for large-scale knowledge exchange and convergence [33]. The chapter by Merelli et al. [34] addresses highperformance computing (HPC) bioinformatics solutions, big data analysis paradigms for computational biology, and the challenges that are still accessible in the fields of healthcare. In particular, the authors pointed out that, thanks to virtualization that prevents transferring too much big data, cloud computing solves big data management and analysis problems in many fields of healthcare. Also, in the particular area of telemedicine, it is critical to have an infrastructure to support high throughput, high capacity of storage, and safe connectivity to allow effective management and automatic analysis of broader patient populations. Horizontal scalability (i.e., the capacity of a device to efficiently extend its resource pool for managing heavy loads) and spatial usability (i.e., capacity to retain performance, usefulness, or usefulness independent of local area concentration advancement to a more dispersed geographic pattern) are two criteria that can be fulfilled by cloud computing [33]. In the medical imaging region, the amount of data can exceed petabytes thanks to high-resolution imaging instruments. It is also apparent that the Cloud Computing Paradigm will render a significant benefit to addressing the computational needs relevant to medical image reconstruction and processing and to facilitate the large exchange of digital images and also advanced control processing. Cloud computing is a new and progressively evolving field of healthcare improvement. In combination with a pay-per-use model, universal, on-demand access to nearly limitless resources allows for new ways of creating, providing, and utilizing services. In an “OMICS setting,” cloud computing is also used in genomics, proteomics, and molecular medicine computing. Medicine is a collaborative and highly data-intensive endeavor [35]. Advances in the OMICS fields produce large quantities of data to be processed and stored (genomics, proteomics, and the like). The subordinate use of medical data with text or data mining techniques also implies an increasing request for complex, accessible services. These tools are also solitary temporarily used so that stable infrastructure projects are difficult to justify and, alternatively, flexible on-demand services are pursued. To meet these demands, cloud computing seems to be a feasible alternative. Commercial providers such as Amazon and Microsoft pledge to make available at their fingertips hundreds of virtual machines, almost instantly and only they are just wanted for the moment. The benefit of such deals is that they only have to be paid for the setup, scale, and period they are essentially used during these services. Massive medical costs and the maintenance of big data during any disease outbreak require technical advances so that at any time and everywhere, everybody has access

8

J. B. Awotunde et al.

to healthcare services. The development of technology has allowed telehealth to provide online healthcare facilities. For patients that are permitted to travel, for villages in rural zones, and for individuals that do not have access to medical care, remote facilities are useful. The uses for telemedicine include the transmission and storage of medical images, video conference patient counseling, continuing education, and facilities in the electronic healthcare field. Sadly, the use of telemedicine technology is hindered by technical and financial costs [36]. To this effect, studies have given cloud computing that offers, among other things, remote support capability, accessible transparent resources, efficient large Internet connectivity, scalable and resources pooling, robust medical data sharing and processing, and the sharing of big data patient records. Many studies have found that inadequate access to patient information is the explanation for most medical errors especially during infectious diseases outbreak [37]. The cloud-based medical system has been regarded as a possible system to increase openness and reduce the extent of medical errors during disease diagnosis to correct health data [38, 39]. Many medical organizations have also chosen cloud storage to obtain and store broad patient data and maintain their electronic health records systems. Electronic health records have evolved rapidly over the last decade, providing a gorgeous basis for data mining to recognize designs and styles in the big data industry in healthcare. Another common point for exchanging medical data is the interchange of electronic health records. By communicating a common hub, these businesses facilitate healthcare sectors to transmit information rather than maintaining ties with many peer businesses [40]. Cloud computing also offers secure storage and sharing resources that can reduce the amount of local traffic to make organizations agile [41]. By reducing the cost needed for starting up automated medical records, which is lacking in many healthcare segment facilities, will improve the efficiency of the healthcare sector [42]. During a disease outbreak, prescriptions and diagnoses, for instance, can be shared through the cloud over different systems. Therefore, for service enhancement and higher standards, hospitals and doctors exchange patient records. The primary advantages of electronic health record cloud storage are the capacity to exchange patient records with other specialists at home and overseas, the facility to pool data in one location, and the capacity to access files anytime, anywhere. Electronic health record cloud computing enables patients to view, replicate and transfer their secure health records [43]. Regardless of the influences of cloud computing to capture and store large health data, the prime problem is the failure of the network, protection, and privacy of patient information that users, hackers, malware, and so on are exploiting [44, 45]. Figure 1 displayed the applicability of cloud computing in the smart healthcare system. There is an increasing influx of people to urban areas today. Healthcare facilities are one of the most critical characteristics that have a major effect on people arriving in city centers during infectious disease outbreaks globally. Metropolises are therefore financing a digital transition to offer residents healthy environments [46]. On the other hand, because of its huge number, high speed, and high variety, conventional models and methods for full conservational performance assessment

Hybrid Cloud/Fog Environment for Healthcare …

9

Elderly and Middle-aged Patients

Doctor

Family

Pharmacis

Healthcare Cl d

Caregiver

Wearable Devices (Pressure Monitor, Pulse, Oximeter, Blood Glucose, Body Temperature) Physical Activities Service Providers

Authority

Digital Telehealth

Fig. 1 Applications of cloud in smart healthcare system

are threatened by the advent of big data [5]. Also, because of their carbon emissions, conventional ICT systems damage the atmosphere [47]. On the other hand, cloud services are a cost-effective medium for accommodating large-scale infrastructure systems have gained considerable acceptance. The use of cloud computing is, therefore, a significant phase in the green processing process that saves resources and protects the atmosphere. The use of sufficient equipment and cloud space saves the organization’s resources and eliminates the costs related to cooling systems, computers, and central servers. Nevertheless, cloud computing supports renewable computing with energy savings, rendering dangerous articles less harmful [13]. Through using intelligent mobile computers, cloud computing has inspired healthcare specialists to observe the wellbeing of patients at home remotely [27]. Besides, IoT will build a network by leveraging integrated sensors to track the patient’s real-time health status and control the treatment process. The IoT would also play an important role in the development of healthcare for the next generation. Although health monitoring systems for IoT-based patients are popular, observing them outdoor hospital requirements increases the IoT’s cloud computing capabilities for the handling and storing of health data [48].

10

J. B. Awotunde et al.

3 Applications of Fog Computing in Smart Healthcare System Fog computing allows a new range of applications and services by extending the CC model. The fog has the following distinct features: (a) heterogeneity of applications, (b) knowledge location and low latency; (c) broad physical circulation; (d) flexibility; (e) produce the huge number of nodes; (f) wireless access predominant; (g) real-time applications and heavy presence of streaming; (h) heterogeneity of applications. Since the edge of the network of fog is not specifically located, they are highly scalable, hence network services between end devices and conventional cloud data centers. Both fog and cloud building blocks are networking tools, storage, and computation, edge network created various features that make fog a non-trivial cloud extension. The help endpoints come with rich network edge resources give birth to fog computing with low latency applications requirement, with applications like augmented reality, video streaming, and gaming. Fog computing targeted applications with a dispersed implementation that are more centralized on the cloud. For instance, fog plays an active role through proxies and access points in providing moving vehicles with high-quality streaming located along highways and tracks. Large-scale sensor networks are other examples of inherently distributed systems that require fog computing and storage resources to monitor the smart grid and the environment. As evidenced in sensor networks, wide range geo-distribution, and smart grid required a large number of nodes. For applications, communicating directly with the mobile devices, thus, highly supports mobility techniques like the LISP protocol, which decouples host identity from location identity. These mobile applications require a distributed directory system. The applications involve real-time interactions rather than batch processing in small cloud applications. The fog nodes will be deployed in a wide environment since the nodes come in various shapes. Streaming is a good example of fog computing supporting certain services seamlessly and requires the cooperation of various providers. Therefore, the applications must be federated across domains and very necessary to be able to interoperate their components. The fog plays a prominent role in the ingestion and processing of the data close to the source. To prolong the life span of the battery or make it harvest electricity, wireless sensor nodes (WSNs) are very useful and are made for this purpose. The WSNs need low bandwidth, low processing power, small memory, unidirectional sink collector sources to function very well. The tinyOS2 a de facto standard operating system is an example of this class of sensor networks that has basic processing and transmission of the static sink with environmental sensing. The motes have been used to collect environmental data like temperature, amount of rainfall, light intensity, and humidity among others [8]. Figure 2 displayed the fog computing ecosystem in a smart healthcare system. The WSNs progressed in many directions like in the areas of wireless sinks, multiple

Hybrid Cloud/Fog Environment for Healthcare …

11

Fig. 2 Fog computing in smart healthcare system

sinks, distributed sinks, and to meet the requirements of modern applications, smartphone sensors have been proposed. The WSNs in successive incarnations have provided energy-restricted in use. The actuators are requiring to perform physical acts in applications beyond sensing and monitoring like in deploying sensors, even carry, open, close, move, focus, and target. The actuators provide sensor networks with new dimensions that can monitor either a device or the measurement process itself. Since the controller node to actuators, sensors to sink, the flow of information is not unidirectional, from the sensors to the sink. The issues of possible oscillatory activity and stability are of importance; it then becomes a closed-loop device in a subtler, but critical way. It has become a concern in a quest to provide real-time and rapid response to works on jitter and latency in the smart healthcare system [8]. In Kashi and Sharifi [8], the contributions of actuator networks and wireless sensor to the coordination of WSANs were discussed. The results of the survey shown that WSAN has two networks in one architectural layout (i) the devices for capture patient data and (ii) mobile ad hoc network (MANET). Banka et al. [27] embedded in their work collaborative adaptive sensing of the atmosphere (CASA) with emphasis on new technologies with higher bandwidth needs a collaborative sensing environment.

12

J. B. Awotunde et al.

Zink et al. [11] deployment information of fog computing to increase healthcare services in remote areas which has been proved used by physician and caregivers. The fogs become the appropriate environment for both WSNs and WSANs having features like proximity and position sensitivity, geo-distribution, the hierarchical organization in helping discharge their energy-restricted [8]. Fog computing takes resources to the edge of the network as an extension to cloud computing. This brings CC closer to IoT-based devices collections center, thus assisting end-users to be closer to the real-time services and speed up their services. These decentralized models bring IoT-based devices closer to the edge where data is generated, thus become the main objective of fog computing. The amount of data transmitted to the cloud for processing and analysis is been reduced by the fog layer, hence increase the security one of the main issues in the IoT-based industry [8]. The fog layer provides a meeting point for storage resources, network nodes, and computing are available for proper management for local data capture to deliver fast results and data are readily obtainable. The trade-off between processing performance and power consumption is provided using low-power system-on-chip (SoC) systems. Cloud servers however behalf like advanced analytics and performed the functions of machine learning jobs combing time series generated by a variety of heterogeneous or mixed kinds of items [48]. With the rapid development in wireless technology, IoT can generate a huge amount of medical data, smart devices, and customized enhance services. Such big medical data can be of countless types, which the cloud server needs to store, process, and thereafter analyze [57]. High latency, security problems, and network traffic arise as a result of the handling of big cloud medical data. Fog computing was introduced to reduce the workload of the CC IoT-based system. The fog acts as a flyover among terminal devices and cloud servers and helps in bringing the cloud service closer to the network edge, hence provided secured and refined smart healthcare services. The edge of the network is an ideal place for analyzing instantaneous generated data closer to where data is created. The feature of fog computing placed it ahead of cloud computing like data pre-processing, local data analytics, data security and privacy, temporary storage, data trimming, distributed, decentralized storage. The combined computing paradigm is needed in an IoT-based applications like health monitoring systems to efficiently perform big data analytics. The use of fog computing has created a better way of handling cloud database for a smarter healthcare system [57].

4 Challenges of Cloud and Fog Computing in Smart Healthcare System The big data 5Vs data importance has a result of the huge amount of patients’ data receives from the medical device such as volume, veracity, variety, and velocity. As a result, a fog node is needed to be connected to receive, store, process, and

Hybrid Cloud/Fog Environment for Healthcare …

13

communicates with IoT-based devices. The system administrative configuration must be controlled to forestall the data fluctuation between fog and cloud computing. To handle various types of data like text, videos, audios, and image files, fog layer regular proprieties with data format are needed from various like smartphone, and a smartwatch. For regular data transfer and urgent data requests, a gateway sufficient router is needed. The portable devices and sensors are used for patient data collection. There is a need for adequate protection for medical facilities so that patient can use their smartphone for health status updates. The use of a smart healthcare system creates possible ways of expanding the healthcare system to the whole population. The appointment time used by patients to see physicians or waits for diagnosis outcomes can be reduced with the use of the intelligent healthcare system. This also provides direct access to real-time medical care and services. To maintain trust between patients and medical experts, the scalability of a smart healthcare system must be taking with all seriousness, and this will in turn save quality time. It is the main concern and it is not appropriate to obtain information from endusers through unauthorized entities, and this, also, poses threats to the personal safety of medical data. The main problem in the introduction and deployment of the smart healthcare system is security and privacy. However, with the integration of these layers, security is needed in any layer, such as the system layer, fog layer, and cloud layer [8]. Another problem facing fog computing is heterogeneity that refers to the various communication capable devices. The devices like smartphones, autonomous vehicles with other IoT smart devices are at the bottom of the layers within the IoT-based system. The heterogeneity within IoT-based systems arises during data processing, data formatting, and data clearing, thus creates difficulty in processing medical information. The enabling of a network to connect with various sensors is an example challenge in a smart healthcare system for the monitoring of patients. For this to take place, heterogeneity must be present when the data is transferred to another system for processing or analysis. The fog layer while going up to the next layer involves various nodes, clusters, switches, and other devices that are needed during data processing and communication facilities [49]. In the designing of architecture that enables multiple monitoring of devices, heterogeneity is, therefore, an important factor to communicate with end-users using IoT-based devices and sensors [11]. There is another research issue in fog computing due to the location of the network, their protection can create a concern. The threats that are not present in an organized cloud architecture arise in fog computing, and this poses a threat in IoT-based system. The major threat from this problem is when an attacker transmits and changes contact between two parties, patients are in the middle of attack [21]. There may be compromise within the gateway between the sensors and fog node in the smart healthcare system. A serious problem may arise if the intruder changed the data being handled by an IoT-based system, this may create serious implications for the welfare of these patients [50]. There is no standard guide and regulations for computing the protocols and interfaces for various products and services in a smart healthcare system. There should

14

J. B. Awotunde et al.

be a standardization put in place to standardize the healthcare system like a dedicated agency is needed to solve this problem. This will help in data dissimilarity and helps to accomplish the real-time response. For good standardization issues like communication protocol, data aggregation interfaces, system interfaces, and gateway interfaces should seriously be considered [51]. There is a broad range of communication protocols, such as WiFi, similar to the application diversity issue. To be interoperable, the fog node should therefore perform the required protocol translation on various core layers. Also, there should be complex regulatory systems before healthcare tools and equipment are available on the market for consumers used. The stakeholders and end-users of e-health products should be contacted by developers for their input on their likes, dislikes, and comforts. This will help to build an acceptable system, interfaces that integrate patient-centric intelligent medical devices and user-friendly interfaces [51]. Authentication techniques that are also needed in cloud–fog computing are one of the security services issues, and it is needed in their architectural design. There must be proper authentication from the end-users or devices before accessing any services to the receiver end like the server. Since fog devices provided services after the end-user authentication, fog computing plays an important role; similarly, both enduser and fog devices must also authenticate by the cloud server before approving access to resources [52]. The one-way authentication approach is too simple for a smart healthcare system. However, due to the vast amount of networks, shared authentication is also equally necessary to avoid cloud masquerade attacks. The security issues were extremely important to be taken into consideration to prevent an attacker that wants to impersonate or breaking the sharing authentication protocol. The fog server will suffer if an attacker gains access to the information of the legitimate end-user, thus, stop end-users from receiving an original message. Creating a safe contact within the fog node, end-users, and cloud servers is another important feature in cloud and fog models. There can be several security issues when an adversary launches an attack. Therefore, this becomes a difficult research challenge in the cloud–fog model to develop a robust authentication protocol [52]. One of the security issues in the three-layer architecture cloud–fog computing paradigm is the man-in-the-middle attack. The attackers trace from the end of the source and modify it before sending another message to the recipient by crack the protection framework. This problem was common to various applications like a wireless sensor network, cloud computing, smart grid among others. Since an attack can be launch within fog and cloud servers, protecting the system became an important security issue [52]. The replay retrieves a message from the sender by the attacker is forwarded to the recipient to initiate an attack. No useful cryptographic contrivance at a standstill that can secure the reply attack. The timestamp technique is one of the current ways to preserve it. However, due to time synchronization issues, it is not ideal for distributed environments. To protect against reply attack, security protection is therefore necessary. In all cryptographic protocols, the perfect forward secrecy property is extremely significant. What if the entity’s private key is somehow revealed, the encrypted key previously identified must not be disclosed to the attacker.

Hybrid Cloud/Fog Environment for Healthcare …

15

When the receiver end receives the message the same as it is sent by the sender, then the principle of data integrity been confirmed and established. It is also one of the significant characteristics required in cloud–fog models. The credibility of the data needs to be provided whenever data needed is been sent from the fog or cloud server. In the same way, integrity should be maintained either to the fog or the cloud server when end-users want to send data. The current cryptography techniques called the hash function like SHA-1, MD5, SHA-2 can provide security and integrity features when used for safety purposes. Numerous investigators have been working on it and are seeking to gain solid assets of honesty. The issues of credibility and effective handling of complexity have remained open research problems in the cloud–fog computing model.

5 The Future Prospects of Cloud and Fog Computing Fog computing is predicted to play a prominent role in Tactile Internet, an emerging technology. The current Internet allows the delivery of content like text, voice, and video, and the skill sets are distributed over the networks through the Tactile Internet. The distribution of the skill set will be achieved by haptic communication, implying the remote control of physical tactile sensations in real time. Telesurgery, telerehabilitation, platoons of cars, and virtual reality are some examples of future applications [11]. The Tactile Internet operates over tactile-human interfaces with a core and edge domain together with a functional architecture that comprises a likely distributed master domain hosted by intelligent Tactile Support Engines. The fog application managed edge works like remote-controlled robots [53]. The major pre-requisite for effective connection in fog computing is ultra-responsive and ultra-reliable connectivity. An end-to-end latency of 1 ms or less is needed and a maximum of one second of outrage yearly. The 5G technology has been identified as an enabler of the Tactile Internet for offering ultra-responsive and ultra-reliable connectivity. The edge and cloud computing are also an enabler of the said technology [53, 54]. The holistic approach of fog computing incorporates end-users and IoT-based devices with fog layer and cloud layer with associated interactions make fog an ideal enabler. The fog systems can naturally be mapped into the Tactile Internet functional architecture. The IoT and end-user layer are part of the managed domain like remotely controlled robots and with a touch, a human–systems interface called the master domains. The managed domains belong to the layer of end-user devices domains while the master domains belong to the IoT-based device’s domain. The edges are the fog layer with the intelligent Tactile Support engines. The cloud can be very useful when there is a need for large storage and powerful processing. Bring the architectural and algorithmic together in fog computer are the prospect of Tactile Internet but their research directions and challenges are beyond the scope of this chapter.

16

J. B. Awotunde et al.

From the architectural perspective, the design of an ultra-responsive and ultrareliable higher layer APIs and protocols for fog device both inter- and intra-layers communications is an instance of the challenge of fog computing in Tactile Internet. Both the transport and application protocols operate on top of the physical and MAC protocols layer required 5G ultra-responsive and ultra-reliable. It will be important to take into account recent attempts to design novel ultra-high data rates [21]. The functionality split between the cloud layer and the fog layer is just another example of a challenge. From the algorithm repositories, the intelligent Tactile Support Engines in the fog layer can be fed into the cloud layer directly. Their components are located in both technologies for efficient storage and processing in the cloud layer and harness mutually proximity in the fog layer. Also, the combination of cloud and fog computing in building a robust IoT-based system will help in edifice a strong and fast processing system in a smart healthcare system. Also, the ultra-responsiveness and ultra-reliability need algorithms running in the device to guarantee touch applications. The fog system provided an enabling platform for running various tasks related to Tactile Internet applications in fogs or cloud layers. There is a need for novel algorithms for task planning due to network traffic that may occur on the computing nodes with other variables where tasks are performed and delay sometimes will far surpass the 1 sm threshold. This is necessary to make sure that activities are carried out within the cumulative threshold and is not surpassed 1 ms. To have optimal task scheduling, novel machine learning and artificial intelligence algorithms are required. To predict behavior and reactions, they may run exclusively in the fog stratum or even in a distributed manner through the cloud and/or fog strata. The fog layer helps to reduce the total network load during network connection in an IoT-based platform, and this will help meet the latency requirement of 1 ms. A various sophisticated algorithms like neural network-based techniques and simple regression models can be considered in modeling the system [55, 56].

6 Conclusion and Future Research Directions The smart healthcare system has the capacity of producing and communicating huge quantities of data produced from IoT-based devices and sensors. The sizes and quantities of networking equipment are more than ever and the volume of data grows even more than ever before. For example, IoT-based devices and sensors can be used to capture and collect patient physiological features and check metrics for patient diagnosis and monitoring. The emergence of cloud computing has touched almost all the human life domains; especially, smart healthcare system has greatly benefited from the paradigm. Cloud computing has brought a technological revolution needed by IoT-based services like high processing, storage capabilities, heterogeneity, and computation resources among others. Nevertheless, in IoT-based systems, cloud computing has weaknesses when it comes to high delays that require real-time response, and it has a low delay response time, thus not match the industrial control

Hybrid Cloud/Fog Environment for Healthcare …

17

system of the smart healthcare system. The fog computing architecture is heterogeneous devices, and geographically distributed ubiquitously connected at the end of a network to provide collaboratively variable and flexible communication, storage services, and computation. The new computing paradigm has various advantages in applications where real time, low latency, and high response time are of utmost importance like in the case of a smart healthcare system. Therefore, the chapter presents a general review of technologies of cloud and fog computing in the smart healthcare system. The applicability of both on smart healthcare systems is discussed; the challenges and the prospects of cloud and fog computing are elaborately discussed. Fog computing presents better infrastructure by providing low latency, distributed processing, better security, fault tolerance, and good privacy when compared with the cloud computing infrastructure. The fog infrastructure sufficiently produces various fog nodes, virtualized data centers, and edge device networks to connect the IoTbased devices to implement large storage and rich cloud computing. Fog computing offered millisecond to sub-second latency faster than real-time interaction, perform better than cloud computing in low-latency applications, and supports multitenancy where cloud computing cannot have provided. In future work, the prospects of both cloud and fog models will be extended in the smart healthcare system to provide a quick and real-time health diagnosis and monitoring for patients suffering from any diseases. Fog computing is very necessary majorly because cloud’s recorded healthcare data may be subject to different types of security risks. Finally, the challenges of cloud and fog computing needed to be looked into and find a lasting solution to ensure reliable and flexible deployment of cloud and fog computing models in smart healthcare systems.

References 1. Kaur, P. D., & Chana, I. (2014). Cloud-based intelligent system for delivering health care as a service. Computer Methods and Programs in Biomedicine, 113(1), 346–359. 2. Fichman, R. G., Kohli, R., & Krishnan, R. (Eds.). (2011). Editorial overview—the role of information systems in healthcare: Current research and future trends. Information Systems Research, 22(3), 419–428. 3. Usak, M., Kubiatko, M., Shabbir, M. S., Viktorovna Dudnik, O., Jermsittiparsert, K., & Rajabion, L. (2020). Health care service delivery based on the Internet of Things: A systematic and comprehensive study. International Journal of Communication Systems, 33(2), e4179. 4. Abimbola, S., Keelan, S., Everett, M., Casburn, K., Mitchell, M., Burchfield, K., & Martiniuk, A. (2019). The medium, the message, and the measure: A theory-driven review on the value of telehealth as a patient-facing digital health innovation. Health Economics Review, 9(1), 21. 5. Adeniyi, E. A., Ogundokun, R. O., & Awotunde, J. B. (2021). IoMT-based wearable body sensors network healthcare monitoring system. In IoT in healthcare and ambient assisted living (pp. 103–121). Springer. 6. Romanow, D., Cho, S., & Straub, D. (2012). Editor’s comments: Riding the wave: Past trends and future directions for health IT research. MIS quarterly, iii–x. 7. Nandyala, C. S., & Kim, H. K. (2016). From cloud to fog and IoT-based real-time U-healthcare monitoring for smart homes and hospitals. International Journal of Smart Home, 10(2), 187– 196.

18

J. B. Awotunde et al.

8. Palumbo, F., Barsocchi, P., Furfari, F., & Ferro, E. (2013). AAL middleware infrastructure for green bed activity monitoring. Journal of Sensors. 9. Saad, M. (2018). Fog computing and its role in the Internet of Things: Concept, security and privacy issues. International Journal of Computers and Applications, 975, 8887. 10. Bonomi, F. (2011, September). Connected vehicles, the internet of things, and fog computing. In The Eighth ACM International Workshop on Vehicular Inter-Networking (VANET), Las Vegas, USA (pp. 13–15). 11. Yadav, A., Kumar Singh, V., Kumar Bhoi, A., Marques, G., Garcia-Zapirain, B., & de la Torre Díez, I. (2020). Wireless body area networks: UWB wearable textile antenna for telemedicine and mobile health systems. Micromachines, 11(6), 558. 12. Awotunde, J. B., Adeniyi, A. E., Ogundokun, R. O., Ajamu, G. J., & Adebayo, P. O. (2021). MIoT-Based Big Data Analytics Architecture, Opportunities and Challenges for Enhanced Telemedicine Systems. Studies in Fuzziness and Soft Computing, 2021, 410, pp. 199–220. 13. Marques, G., Miranda, N., Kumar Bhoi, A., Garcia-Zapirain, B., Hamrioui, S., & de la Torre Díez, I. (2020). Internet of Things and enhanced living environments: Measuring and mapping air quality using cyber-physical systems and mobile computing technologies. Sensors, 20(3), 720. 14. Rajabion, L., Shaltooki, A. A., Taghikhah, M., Ghasemi, A., & Badfar, A. (2019). Healthcare broad data analysis techniques: The role of cloud services. International Journal of Management of Intelligence, 49, 271–289. 15. Ali, O., Shrestha, A., Soar, J., & Wamba, S. F. (2018). Cloud computing-enabled healthcare opportunities, issues, and applications: A systematic review. International Journal of Information Management, 43, 146–158. 16. Palanisamy, V., & Thirunavukarasu, R. (2019). Implications of big data analytics in developing healthcare frameworks—A review. Journal of King Saud University-Computer and Information Sciences, 31(4), 415–425. 17. Aceto, G., Persico, V., & Pescapé, A. (2018). The role of information and communication technologies in healthcare: Taxonomies, perspectives, and challenges. Journal of Network and Computer Applications, 107, 125–154. 18. Dash, S., Shakyawar, S. K., Sharma, M., & Kaushik, S. (2019). Big data in healthcare: Management, analysis, and prospects. Journal of Big Data, 6(1), 54. 19. Ali, O., Soar, J., & Shrestha, A. (2018). Perceived potential for value creation from cloud computing: A study of the Australian regional government sector. Behaviour and Information Technology, 37(12), 1157–1176. 20. Senyo, P. K., Addae, E., & Boateng, R. (2018). Cloud computing research: A review of research themes, frameworks, methods, and future research directions. International Journal of Information Management, 38(1), 128–139. 21. Oniani, S., Marques, G., Barnovi, S., Pires, I. M., & Bhoi, A. K. (2021). Artificial intelligence for Internet of Things and enhanced medical systems. In Bio-inspired neurocomputing (pp. 43–59). Springer. 22. Bayramusta, M., & Nasir, V. A. (2016). A fad or future of IT? A comprehensive literature review on cloud computing research. International Journal of Information Management, 36(4), 635–644. 23. Ahuja, S. P., Mani, S., & Zambrano, J. (2012). A survey of the state of cloud computing in healthcare. Network and Communication Technologies, 1(2), 12. 24. Abdulraheem, M., Awotunde, J. B., Jimoh, R. G., & Oladipo, I. D. (2021). An Efficient Lightweight Cryptographic Algorithm for IoT Security. Communications in Computer and Information Science, 2021, 1350, pp. 41–53. 25. Culyer, A. J., & Chalkidou, K. (2019). Economic evaluation for health investments En route to universal health coverage: Cost-benefit analysis or cost-effectiveness analysis? Value in Health, 22(1), 99–103. 26. Elavarasan, R. M., & Pugazhendhi, R. (2020). Restructured society and environment: A review on potential technological strategies to control the COVID-19 pandemic. Science of the Total Environment, 138858.

Hybrid Cloud/Fog Environment for Healthcare …

19

27. Mora, N., Grossi, F., Russo, D., Barsocchi, P., Hu, R., Brunschwiler, T., Ciampolini, P., et al. (2019). IoT-based home monitoring: supporting practitioners’ assessment by behavioral analysis. Sensors, 19(14), 3238. 28. Venkatesh, V., Rai, A., Sykes, T. A., & Aljafari, R. (2016). Combating infant mortality in rural India: Evidence from a field study of eHealth kiosk implementations. MIS Quarterly, 40(2), 353–380. 29. Koufi, V., Malamateniou, F., & Vassilacopoulos, G. (2010, November). Ubiquitous access to cloud emergency medical services. In Proceedings of the 10th IEEE International Conference on Information Technology and Applications in Biomedicine (pp. 1–4). IEEE. 30. Sharma, M., & Sehrawat, R. (2020). A hybrid multi-criteria decision-making method for cloud adoption: Evidence from the healthcare sector. Technology in Society, 101258. 31. Seth, B., Dalal, S., & Kumar, R. (2019). Securing bioinformatics cloud for big data: budding buzzword or a glance of the future. In Recent advances in computational intelligence (pp. 121– 147). Springer. 32. Aceto, G., Persico, V., & Pescapé, A. (2020). Industry 4.0 and health: Internet of Things, big data, and cloud computing for healthcare 4.0. Journal of Industrial Information Integration, 18, 100129. 33. Calabrese, B., & Cannataro, M. (2015). Cloud computing in healthcare and biomedicine. Scalable Computing: Practice and Experience, 16(1), 1–18. 34. Awotunde, J. B., Jimoh, R. G., Oladipo, I. D., & Abdulraheem, M. (2021). Prediction of Malaria Fever Using Long-Short-Term Memory and Big Data. Communications in Computer and Information Science, 2021, 1350, pp. 41–53. 35. Navale, V., & Bourne, P. E. (2018). Cloud computing applications for biomedical science: A perspective. PLoS Computational Biology, 14(6), e1006144. 36. Jin, Z., & Chen, Y. (2015). Telemedicine in the cloud era: Prospects and challenges. IEEE Pervasive Computing, 14(1), 54–61. 37. Sundwall, D. N., Munger, M. A., Tak, C. R., Walsh, M., & Feehan, M. (2020). Lifetime prevalence and correlates of patient-perceived medical errors experienced in the US ambulatory setting: A population-based study. Health Equity, 4(1), 430–437. 38. Maduravoyal, C. (2018). Patient-controlled personal health record enforcing patient privacy in the cloud-based healthcare system. International Journal of Pure and Applied Mathematics, 119(10), 375–392. 39. Sahoo, P. K., Mohapatra, S. K., & Wu, S. L. (2018). SLA based healthcare big data analysis and computing in a cloud network. Journal of Parallel and Distributed Computing, 119, 121–135. 40. Regola, N., & Chawla, N. V. (2013). Storing and using health data in a virtual private cloud. Journal of Medical Internet Research, 15(3), e63. 41. Abatal, A., Khallouki, H., & Bahaj, M. (2018, April). A semantic smart interconnected healthcare system using ontology and cloud computing. In 2018 4th International Conference on Optimization and Applications (ICOA) (pp. 1–5). IEEE. 42. Schweitzer, E. J. (2012). Reconciliation of the cloud computing model with US federal electronic health record regulations. Journal of the American Medical Informatics Association, 19(2), 161–165. 43. Chen, S. W., Chiang, D. L., Liu, C. H., Chen, T. S., Lai, F., Wang, H., & Wei, W. (2016). Confidentiality protection of digital health records in cloud computing. Journal of Medical Systems, 40(5), 124. 44. Hayes, A. N. (2016). Is cloud computing in healthcare providing a safe environment for storing protected health information? A systematic review and meta-analysis. 45. Mizani, M. A. (2017). Cloud-based computing. In Key advances in clinical informatics (pp. 239–255). Academic Press. 46. Yassine, A., Singh, S., & Alamri, A. (2017). Mining human activity patterns from smart home big data for health care applications. IEEE Access, 5, 13131–13141. 47. Ayo, F. E., Awotunde, J. B., Ogundokun, R. O., Folorunso, S. O., & Adekunle, A. O. (2020). A decision support system for multi-target disease diagnosis: A bioinformatics approach. Heliyon, 6(3), e03657.

20

J. B. Awotunde et al.

48. Ghanavati, S., Abawajy, J. H., Izadi, D., & Alelaiwi, A. A. (2017). Cloud-assisted IoT-based health status monitoring framework. Cluster Computing, 20(2), 1843–1853. 49. Oladipo, I.D., Babatunde, A.O., Awotunde, J.B., Abdulraheem, M. (2021). An Improved Hybridization in the Diagnosis of Diabetes Mellitus Using Selected Computational Intelligence. Communications in Computer and Information Science, 2021, 1350, pp. 272–285. 50. Rehman, H. U., Khan, A., & Habib, U. (2020). Fog computing for bioinformatics applications. In Fog computing: Theory and practice (pp. 529–546). 51. Devarajan, M., Subramaniyaswamy, V., Vijayakumar, V., & Ravi, L. (2019). Fog-assisted personalized healthcare-support system for remote patients with diabetes. Journal of Ambient Intelligence and Humanized Computing, 10(10), 3747–3760. 52. Kunal, S., Saha, A., & Amin, R. (2019). An overview of cloud-fog computing: Architectures, applications with security challenges. Security and Privacy, 2(4), e72. 53. Yi, S., Li, C., & Li, Q. (2015, June). A survey of fog computing: Concepts, applications, and issues. In Proceedings of the 2015 Workshop on Mobile Big Data (pp. 37–42). 54. Mouradian, C., Naboulsi, D., Yangui, S., Glitho, R. H., Morrow, M. J., & Polakos, P. A. (2017). A comprehensive survey on fog computing: State-of-the-art and research challenges. IEEE Communications Surveys and Tutorials, 20(1), 416–464. 55. Al-Khafajiy, M., Webster, L., Baker, T., & Waraich, A. (2018, June). Towards fog driven IoT healthcare: Challenges and framework of fog computing in healthcare. In Proceedings of the 2nd International Conference on Future Networks and Distributed Systems (pp. 1–7). 56. Kumari, A., Tanwar, S., Tyagi, S., & Kumar, N. (2018). Fog computing for Healthcare 4.0 environment: Opportunities and challenges. Computers and Electrical Engineering, 72, 1–13. 57. Ayo, F. E., Ogundokun, R. O., Awotunde, J. B., Adebiyi, M. O., & Adeniyi, A. E. (2020, July). Severe acne skin disease: A fuzzy-based method for diagnosis. In Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics) (Vol. 12254, pp. 320–334). LNCS. 58. Puthal, D., Obaidat, M. S., Nanda, P., Prasad, M., Mohanty, S. P., & Zomaya, A. Y. (2018). Secure and sustainable load balancing of edge data centers in fog computing. IEEE Communications Magazine, 56(5), 60–65. 59. Simsek, M., Aijaz, A., Dohler, M., Sachs, J., & Fettweis, G. (2016). 5G-enabled tactile internet. IEEE Journal on Selected Areas in Communications, 34(3), 460–473. 60. Maier, M., Chowdhury, M., Rimal, B. P., & Van, D. P. (2016). The tactile internet: Vision, recent progress, and open challenges. IEEE Communications Magazine, 54(5), 138–145. 61. Petrov, I., & Janevski, T. (2016, October). Design of novel 5G transport protocol. In 2016 International Conference on Wireless Networks and Mobile Communications (WINCOM) (pp. 29–33). IEEE. 62. Sakr, N., Georganas, N. D., Zhao, J., & Shen, X. (2007, July). Motion and force prediction in haptic media. In 2007 IEEE International Conference on Multimedia and Expo (pp. 2242– 2245). IEEE. 63. Marques, G., Bhoi, A. K., de Albuquerque, V. H. C., K. S., H. (Eds.). (2021). IoT in healthcare and ambient assisted living. Springer.

Hybrid Intelligent System for Medical Diagnosis in Health Care Moolchand Sharma, Akanksha Kochhar, Deepak Gupta, and Jafar Al Zubi

Abstract The autonomous city implies a global vision that incorporates artificial intelligence, deep learning, big data, decision-making, ICT and the Internet of Things (IoT) to promote sustainable development. The ageing issue is that researchers, companies and the government should devote efforts to developing smart health care, innovative technology and applications. For an extended period, conventional intelligence systems have played a critical role in health care. However, with the increased popularity and widespread use of these hybrid intelligent computer systems, there is a significant shift in the healthcare sector. Diagnosis and detection of various diseases using several techniques can be resolved by using these techniques. Different novel methods are applied to biomedical engineering to diagnose diseases, and new models are being studied and compared with the existing technologies. Hybrid intelligence systems can be implied in decision-making, remote monitoring, healthcare logistics, medical diagnosis, and modern information system. This success’s fundamental cause seems to be derived by different intelligent computational mechanisms, such as genetic algorithms, evolutionary computation, convolutional neural network (CNN), long short-term memory (LSTM), autoencoders, deep generative models and deep belief networks. To solve complex problems, we need domain knowledge that comprises the methodologies that provide hybrid systems with complementary reasoning and empirical data. This chapter will focus on the need for a hybrid intelligent system in the healthcare industry and their medical diagnosis applicability.

M. Sharma (B) · A. Kochhar · D. Gupta Maharaja Agrasen Institute of Technology, MAIT, Delhi (GGSIPU), India e-mail: [email protected] A. Kochhar e-mail: [email protected] D. Gupta e-mail: [email protected] J. A. Zubi School of Engineering, Al-Balqa Applied University, Jordan, India e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 A. Kumar Bhoi et al. (eds.), Hybrid Artificial Intelligence and IoT in Healthcare, Intelligent Systems Reference Library 209, https://doi.org/10.1007/978-981-16-2972-3_2

21

22

M. Sharma et al.

Keywords Intelligent system · Deep learning · Genetic algorithms · Neural networks · Hybrid intelligence system · Computational intelligence

1 Introduction For medical applications and intelligent computerized system, research is a significant and thrilling area. Based on the symptoms, a general doctor accumulates typically his/her insight dependent on patients’ history and affirms by diagnosing them. Thus, a physician’s experience is vital for diagnostic accuracy and predictive importance of symptoms towards certain diseases. As rapid strides have been made in the medical field and therapy of that particular cause, e.g. the evolution of new diseases and drug accessibility, it has become stimulating/demanding for a doctor to stay updated with all the current knowledge developments clinical practice [1]. Additionally, with the development of computer technologies, it is easy to gather and store a massive amount of information in the digital form, e.g. an exclusive database of patient records in electronic form [2]. In such a scenario, applying an electronic medical decision support system is a feasible approach that would assist physicians so that patients’ diagnosis can be made quickly and efficiently [3]. However, several problems have to be solved before an efficient/effective medical support system, i.e. decision support system (DSS), can be formulated and applied. It includes resultmaking in the presence of multiple factors, like inaccuracy and uncertainty [4]. Though it is important in terms of expert’s knowledge and experience, which ranges from the assessment of a patient’s condition so that a diagnosis can be made, developments in machine learning techniques have resulted in paving the way for medicinal physicians to use computerized hybrid systems in their field, e.g. medical imagery and X-ray photography [5]. A physician must conclude the expected disease to the root cause among the list of possible causes with similar symptoms by applying his/her experience and skill and then confirming the disease using multiple tests [6]. Simultaneously, hi-tech intelligent systems can help physicians make rational and well-informed inferences quickly, e.g. by imbibing past experiences from electronic patient records and making an inference regarding the diagnosis of a present patient proper rationale. The advantage of using intelligent systems is accurate in detecting, diagnosing, and diminishing time and expenses related to patient treatment [7]. The primary objective of developing/formulating machine learning models is to support various decision-making tasks in medicine. For example, intelligent classifiers are used for diagnosis, prognosis and airing of diabetes, breast cancer and Parkinson’s disease [8]. The capability of fuzzy neural models to learn from the facts (i.e. patient records) and generalizing the process beyond the training samples makes them perfect for classifiers’ task to detect any disease [9]. It includes fuzzy learning vector quantization networks, vague neural networks and vague probabilistic neural networks [10]. However, they are hit by a limitation, i.e. lack of these models’ ability to explain their predictions [11]. This is the inspiration for this study. We try to invent

Hybrid Intelligent System for Medical Diagnosis in Health Care

23

the machine knowledge-based system that can reflect its reasoning for handling an input case and providing the rationale behind its forecasts. This chapter has discussed the hybrid intelligent model, which has two real-world influences in medicinal decision support. First, the capability to provide rationale and validation for prediction is the most important to convince medical practitioners that computerized decision support can be efficient. This skill is of prime importance in security applications, such as medical diagnosis and prognosis, whereby practitioners need to understand and form a conclusion regarding the conclusion-forming process used by a computerized system to make accurate predictions [12]. It can become a source of second opinions in situations involving medical diagnosis. The rules in the form of a decision tree from the hybrid model are of paramount importance in practice [13]. Secondly, it is critical for a decision support system to have a high accuracy in medical applications. As stated before, an elevated false-negative rate would result in increased risk for patients as they would be deprived of obtaining essential facilities medical. In contrast, a highly inaccurate alarm rate would result in unnecessary stress and anxiety in patients and increased medical resources pressure due to huge demand. Nevertheless, a DSS with the right positive rate and accurate negative rate results in lower therapeutic costs and vestibular patients’ development in an otolaryngology clinic [14]. Also, it is a fact that machine learning models are useful in minimizing cost and time for medical diagnosis [15].

1.1 Intelligent Systems What is Intelligence? A machine can learn from training data, compute, reason, perceive relationships and analogies, store and retrieve information from memory, solve problems, comprehend complex ideas, use the natural language fluently, classify, generalize, and adapt to new situations. Types of Intelligence A.

Linguistic Intelligence:

The capability to talk, recognize and use mechanisms of descriptive linguistics (speech sounds), syntax (grammar) and linguistics (meaning), e.g. Narrators, Orators. B.

Musical Intelligence:

The capability to make, communicate with and perceive meanings made from sound, understanding of pitch, rhythm, e.g. Musicians, Singers, Composers. C.

Logical-mathematical Intelligence:

24

M. Sharma et al.

The capability to use and perceive relationships within the absence of action or objects. Understanding advanced and abstract concepts, e.g. Geometricians, Researchers. D.

Spatial Intelligence:

The capability to understand visual or spacial info, modify it, recreate visual pictures, construct 3D pictures, and manoeuvre and rotate them, e.g. Map readers, Astronauts, Physicists. E.

Bodily Kinesthetic Intelligence:

The willingness to use a whole or part of the body to solve problems or fashion items, handle fine and coarse motor skills, and control objects, e.g. Singers, Players. F.

Intra-personal Intelligence:

The capability to distinguish among one’s feelings, intentions and motivations, e.g. Gautam Buddha. G.

Interpersonal Intelligence:

The capability to identify and make distinctions among other people’s feelings, beliefs and intentions, e.g. Mass Communicators, Interviewers. We should know at least what does intelligence composed of? The intelligence is impalpable. It is composed of different attributes as mentioned below and also shown in Fig. 1. i. ii. iii. iv.

Intellectual, i.e. reasoning Knowledge Problem-solving Observation, i.e. perception

Fig. 1 Classification of intelligence

Hybrid Intelligent System for Medical Diagnosis in Health Care

v.

25

Linguistic intelligence. Let us quickly go through all the components.

A. B.

Intellectual—Two types permit us to provide the basis for judgement, making decision and prediction. Knowledge—By learning, practising, being taught or experiencing something, we can acquire knowledge and skills.

Knowledge improves the consciousness of the themes of the study. The capability of knowledge is possessed by different categories like humans, some animals and AI-enabled systems. Learning/knowledge is categorized as— i. ii. iii. iv. v.

vi.

vii.

viii.

C.

E.

Sensory System Knowledge—Students are listening to recorded audio lectures using listening and hearing. Episodic Knowledge—To learn by remembering the order of events that you have heard or seen. It is ordered and linear. Motor Education—It is the method of muscles’ precise movement, for example, object picking, writing, etc. Data-Based Knowledge—It is the process of watching others and receiving information. The infant, for instance, tries to learn by imitating its parent. Sensory Activity Knowledge—It is the mechanism by which stimuli that one has seen before can be recognized. They are identifying, for example, and classifying objects and circumstances. Relational learning requires learning, rather than simple properties, to distinguish between various stimuli based on relational properties. For Example, adding a very small amount of salt while cooking potatoes that came up salty last time, when cooked with adding, say, a tablespoon of salt. Abstraction Learning—Via visual stimuli like pictures, colours, maps, etc., it is learning. An individual may build a roadmap in mind, for instance, before actually following the route. Stimulus–Response Learning—When a particular stimulus is present, it is the process of learning to conduct specific behaviour. For example, a dog becomes cautious by raising its ear when it hears the doorbell. Problem-Solving—Based on past movements, it is the phase in which one sees and observes can attempt to overcome the present situation, blocked by known or unknown barriers. Problem-solving often requires decision-making, choosing the most appropriate option from various solutions to achieve the ultimate goal. Observation—It is the method of sensory knowledge processing, perception, collection and organization. Perception presumes sensitivity—sensory organs of humans aid vision. The sensor method incorporates the data obtained by the sensors in a practical way within the AI domain. It is the method of sensory knowledge processing, perception, collection and organization. Perception presumes sensitivity—sensory organs of humans aid vision. The

26

F.

M. Sharma et al.

sensor method incorporates the data obtained by the sensors in a practical way within the AI domain. Philological Intelligence—The study of oral language and the spoken and written language of speech and writing. In interpersonal communication, it is important.

1.2 Hybrid Intelligent System Intelligent systems are hybrid and are considered the most important research field for computing intelligence in modern ways and only concerned with developing intelligent systems for the next generation. To investigate the hybrid systems, a stimulus must be the spread of awareness in the communities leading with academics which combines different methods to solve the problems related to artificial intelligence (AI). The combination of different learning methods and adaption is required to overcome the drawbacks to achieve synergetic effects with a combination of two or more than two computational methods or inclusion of techniques that have contributed to the intelligent systems that are newly designed. These approaches have followed an ad hoc method which may not certainly be used. Different frameworks are made for modelling expertise in the current ages: one is soft computing approaches, support of decisions, segmentation techniques used for images and videos, process control, robotics, automation, etc [16]. Most of the techniques use various information acquisition schemes, models for decision-making, and learning strategies to solve the computation problem. This helps to overcome the drawbacks of one is through more than one methods or a combination of different methods. These notions lead to different emerging architectures of intelligent systems [17]. We are familiar with the fact that intellectual systems provide human-like information such as domain data, indeterminate reasoning, and a noisy and time-variable environment to handle computing problems. The most important phase is designing the hybrid systems as the main aim is to combine and interact with various techniques and emerge with a new working technique. The properly known methods can be applied to a specific problem, and that problem can be solved within the system. If a system is faced with some problem, then the drawbacks must be addressed with traditional systems. So, to develop a hybrid system will lead to an evolution in the traditional systems. Hybrid intelligent architecture classification is carried out in four separate groups depending on the system’s overall success, namely stand-alone, transformative, hierarchical hybrid and integrated hybrid system [18]. Fused architectures are the first proper type of interconnected intelligent systems. They have structures that combine a single machine model with various techniques. They swap systems with data and information representations. Another method is to place the different strategies sideby-side in a problem-solving task and rely on their interaction. Robustness, higher performance and increased problem-solving skills gain from integrated models. Finally, a broad range of features, including adaptation, widespread use, noise tolerance and rationale, can be supported by fully integrated models. The two architectures

Hybrid Intelligent System for Medical Diagnosis in Health Care

27

belong to the integrated model in this chapter. Combining neural network learning with evolutionary computation, Sect. 1.2.1 presents the hybrid mechanism for the inference system’s refinement. In this section, an example of an application is also given. We also addressed a hybrid compound of the fuzzy clustering algorithm in Sect. 1.2.2 and a fuzzy inference method for a web mining task.

1.2.1

Adaption of Fuzzy Inference Systems

In traditional times, the fuzzy inference method uses the expert model, which defines the process’s essential properties. Expert expertise is the key source of the design of the fuzzy inference system. If the performance indicator is taken into account, it is appropriate to adopt the following, such as the membership function, the knowledge base, the inference process. It takes a lot of research work to adapt to the fuzzy inference method [19]. It involves the adaptation of the functions of membership, rule bases and aggregation operators. They contain but are not restricted to: • It is well known that the self-organizing fuzzy controller proposed by Procyk considers the problem of rule generation and adaptation [20]. • The gradient descent and the variants applied to alter the member functions’ input and output parameters [21]. • Cutting the amount and adjusting the state of input/output participation capacities [22]. • Apparatuses to distinguish the models of fuzzy models. • The fuzzy principles’ hypothesis generally applies to the min and max operators for fuzzy intersections and union. If T-norm and T-conorm operators are parameters, the gradient descent technique can fine-tune the fuzzy operators in a supervised learning setting. The local fuzzy region defines the antecedent of fuzzy rule, while a consequent shows the region’s behaviour through various constituents. It comprises a linear equation or a membership function [23]. The fuzzy inference systems are adapted using techniques that are evolutionarily computed and have been explored widely. This adaption of relationship function is known as self-tuning. Adaptation of fuzzy derivation frameworks utilizing transformative calculation strategies has been broadly investigated [24]. The programmed adjustment of participation capacities is prevalently called self-tuning. The genome encodes parameters of the trapezoidal, triangle, calculated, hyperbolic-digression, Gaussian enrollment works, etc [25, 26]. Fuzzy rules by using the evolutionary search can be performed using three approaches. • The foremost approach is the Michigan approach, where the fuzzy knowledge base is considered, and it results in the antagonistic rules to lead competition and cooperation with the fuzzy rules. To represent a fuzzy rule, the use of genotype is done, and the solution is considered out of the entire population.

28

M. Sharma et al.

• Another method could be the Pittsburgh approach which evolves knowledge from the population instead of applying fuzzy rules to the individual. The new collaboration of rules is done in order to serve the genetic operators. It may suffer from a drawback: the complexity is increased of search space, and the computation burden is also overhead, mostly when online learning is done. • The third method is the iterative knowledge approach, similar to the primary method, where every chromosome represents a single rule. However, in divergence, to the first approach, only the chromosome which is superlative will form the solution, and the left ones are discarded [27]. The developmental learning process has developed to finish the standard base using an iterative learning process. Utilizing the neuro-fuzzy model, no affirmation is given that the neural system calculation will join, and for the fuzzy framework to be fruitful, the tuning is to be done. It incorporates the assurance of ideal parameter estimations of the participation capacities, procedure on fuzzy, etc. One of the developmental frameworks’ kind elements is that they are versatile to any condition [28]. The test confirmations demonstrate that developmental calculations are wasteful for tuning arrangements. However, they are viewed as proficient when finding global basins of attraction [27, 29]. The efficacy in evolutionary is enhanced if a limited search process is considered. These calculations first discover an area in the space and afterwards take care of the nearby looking through a technique to locate an ideal arrangement. To locate a decent beginning parameter esteem by finding the great locale in space is truly fascinating. To characterize the bowl of fascination for the neighbourhood least worth includes all the focuses, parameter esteems that may join to a nearby least using a calculation lastly finding a worldwide least incentive with the assistance of a neighbourhood search calculation. Referring to Fig. 2, tt1 and tt2 are considered the original parameter values cited by the evolutionary algorithms, and WA and WB correspond to the concluding parameter that tweaks the technique of meta-learning. Now the model which we present is the evolving neural fuzzy model. The function of this model is that it optimizes the inference system of fuzzy by using a metaheuristic approach that combines the neural Fig. 2 Fine-tuning using hybrid learning parameters

Hybrid Intelligent System for Medical Diagnosis in Health Care

29

Fig. 3 EVoNF general computational framework

networks with evolutionary computation. The proposed technique is a methodology that can integrate various neural networks and fuzzy inference systems, along with the evolutionary search process [24, 30]. Figure 2 shows the connection component with the evolutionary search, which advances at the best level with the time scale at the slowest pace. For each developmental search calculation that supports fuzzy operation, it requires the quest for the fuzzy rule. It requires a short timescale to choose the induction framework and the issue proclamation. Therefore, the fuzzy interpretation systems have evolved at a slow rate, whereas the evolution of the quantity and the membership type is evolving faster. Other layers also provide some functions. The adaptation layers form a hierarchy that depends on previous knowledge. If there is complete knowledge about the knowledge base, then the inference mechanism will be better implemented and acceptable. If some fuzzy inference system is considered best to solve a problem, then the computational task will be minimized as the search space will be reduced. The chromosome architecture is shown in Fig. 3. Figure 4 shows the engineering architecture, and the component that evolves can be considered by way of a basis for the adaptive fuzzy systems. Every layer of the hierarchical evolutionary search procedure is characterized by a chromosome used for efficient modelling of EVoNF. The comprehensive working and demonstrating procedure is considered by way of: Layer 1: It is the easiest way to encode each input variable’s membership functions and parameters for every membership function. Figure 5 shown represents the chromosome of a membership function represented by the parameters represented by the variables. The minimal constraints of membership functions are located in the evolutionary algorithm, which is tuned by the network’s flow process. To represent a Mamdani fuzzy inference system, a similar approach is used to output the membership function. The master’s recommendation is to evaluate the MF shape and its structures parameters to gauge the hunt space. The precise coding

30

M. Sharma et al.

Fig. 4 Chromosome structure of EVoNF model

Fig. 5 Representation of n membership functions of a bell-shaped MF

strategy proposed by Cord’on is considered to speak to the resulting parameters of the Takagi–Sugeno derivation framework [19]. Layer 2: In this layer, the optimization of the rule base is performed, which comprises the overall sum of rules, the representations of the parts that is the antecedent and the consequent. The rules increase drastically, increasing the number of variables and the fuzzy sets based on the representation. The easiest method is representing a gene that characterizes a single rule where ‘1’ is used for selected, and ‘0’ is used for rejection. Figure 6, shown below, represents the structure of a chromosome. To show the single rule, a position-dependent code is used with all the elements by way of the system’s number of variables. Each component is a parallel string that is a piece in the fuzzy set, which shows the

Fig. 6 Representation of the entire rule base consisting of m fuzzy rules

Hybrid Intelligent System for Medical Diagnosis in Health Care

31

Fig. 7 Depiction of an individual fuzzy rule

nonappearance or nearness of the relating semantic name utilized in the standard. On the off chance that we think about three information and one yield variable, with fluffy allotments, contains 3, 2, 2 fluffy sets for input factors and three fuzzy sets for yield variable, the fuzzy guideline will have an illustration as presented in Fig. 7. Layer 3: This layer represents a chromosome with a number of the same T-norm and T-conorm operator parameters. Representation in real numbers is considered appropriate to characterize the fuzzy operator parameters. These parameters are refined by techniques called gradient descent. Layer 4: The responsibility of this layer is to select the minimum learning parameters. The gradient descent strategies are known for their exhibition that relies legitimately upon the learning rate whenever contrasted with the error surface. Representation in real numbers is used for the representation of learning parameters. The evolutionary algorithms decide the learning parameters, which are further used to refine the membership functions and the method used for inference rules. Layer 5: Interaction with the environment is done in this layer which further decides the inference system and is considered minimal according to the environment. When the representation of the chromosome is completed in the EVoNF model, the evolutionary search procedure can be started as follows: Step 1: Initially, the initial population’s generation is done in N no’s of C chromosomes. The fitness function is calculated for every chromosome, but it entirely depends on the problem. Step 2: Once the fitness function is calculated, the selection method is used to produce the children for every individual from the current population. Step 3: The genetic operators are applied to every individual from the populace to generate the next generation. Step 4: If the present model is completed, then the error rate is calculated. Step 5: End.

1.3 Health Care Through prevention, diagnosis and treatment of illness, sickness, impairment and other physical and mental conditions, health care means maintaining or improving human health. Allied health professionals provide health treatment. Physicians

32

M. Sharma et al.

and physician assistants comprise these health professionals. Dentistry, midwifery, nursing, medicine, optometry, audiology, pharmacy, counselling, occupational therapy, physical therapy and other healthcare professions are included in all healthcare systems. This includes jobs in primary care, secondary care, tertiary care and public health sectors. Access to health care, mainly influenced by social, economic and health policies, can vary across nations, communities and individuals. ‘The use of personal health care on time for the best health outcomes’ [31] means the availability of health services. There are hardly any elements that should be considered to enter the human services offices, including monetary restrictions such as security inclusion and geographical barriers (travel costs). The possibility of taking care of the time of work to use administrations and individual impediments such as the inability to speak to social insurance companies, unforeseen health treatment and treatments’ efficacy. The overall performance is adversely affected by healthcare constraints (well-being, mortality rates). Health facilities are businesses that are structured to meet individual communities’ health needs. According to the WHO, a well-functioning healthcare system requires a financing facility, well-trained and adequately paid personnel, sound policy and decision-making expertise, and well-maintained health facilities to provide quality medicines and technology.

2 Need for Health Care’s Intelligent Infrastructure for Medical Diagnosis The development in the field of intelligent systems has been improved. It has shown excellent results in ability and reliability by using different techniques for computersupported medical diagnosis. The fundamental justification for using the machine is this. In massive disease prevention, including cancers: breast cancer, melanoma, etc., medical treatment is only successful if the disease is diagnosed in the early stages [32]. Many different technologies are now available, and the question is how many laboratory examinations will diagnose them. New steps can be taken towards a computer-supported medical diagnosis to implement learning methods based on various mathematical theories. This can be beneficial, particularly in situations where there are inadequate experts, and the critical element is rapid diagnosis. This chapter describes an innovative method of classification capable of handling cases with limited data sets and nonlinear reciprocal relationships.

2.1 Basic Algorithm A small portion of training data called support vectors (SV) is the algorithm’s concept, which is similar to an optimal separation of the entire data set. Suppose optimum separation means that the minimum distance between the two groups for the closest

Hybrid Intelligent System for Medical Diagnosis in Health Care

33

point is maximum from the separating hyperplane. If the hyperplane’s function is y = m Xi + c, then finding the optimal hyperplane means minimizing ||m||2 . For nonlinear tasks such as regression, recognition, regulation, fluid system modelling, the nonlinear support vector approach is used. To map the data to a higher space where linear regression is feasible when nonlinear mapping is used. When the kernel functions are introduced, this change is possible. They come from the Hilbert Spaces kernel replication theory [33, 34].

2.2 Applications in Diagnosis The wrong diagnosis can lead to a significant threat to the quality and safety of health care. The rates of ambulatory diagnostic errors are projected to be 5.08% in the USA, equivalent to 12 million adults per year. Approximately, fifty percent of these errors may be detrimental. Because of the high volume of medical image data, hybrid technology was used to improve medical diagnosis quality, particularly in radiology [35]. Keith Dreyer, a radiologist at the Harvard Medical School, said that “Meaningful technology would enhance consistency, productivity, and performance”. Esteva et al. trained deep convolutionary neural networks (CNNs) based on 129,450 clinical images for skin cancer diagnostics. The findings show that this method can classify skin cancer at a level comparable to dermatologists. You thought that smartphones could be a cheap way to help dermatologists broaden their scope to improve their access to diagnostic treatment [36]. Liu of Google, Inc. has announced the CNN framework for the pathological detection of lymph node breast cancer metastasis. The findings showed that this technology would improve the tempo, precision and uniformity of diagnosis and decrease the false-negative rate to one-fourth of that for human pathologists. AI technology has been extensively adopted into numerous areas of diagnosis. For instance, recently published a cardiac motion magnetic resonance imaging algorithm to forecast patients’ pulmonary hypertension outcomes accurately. In the study of electrocardiograms in chronically ill patients, Moss et al. employed an automatic rhythm classification approach. They concluded that AI technology provided new information and data observations that physicists could have ignored. Lee et al. have also addressed promising outcomes from recent studies using AI in stroke imaging and suggested that AI technology may play a key role in stroke patient treated with an individualized strategy. Current health systems focus on treatment-based medicine and do not provide safe patients at high risk of reliable, low-cost treatments. At the same time, a vast global economic burden has been created by the pandemic. In the USA, dissatisfaction with the healthcare system has prompted the government to consider incorporating AI personal health monitoring platforms in healthcare management systems, decreasing healthcare costs and strengthening health quality. AiCure, a mobile application funded by the United States National Institutes of Health, was created to

34

M. Sharma et al.

monitor patients’ medications and conditions. Because of advancements in insensitivity, over-use in health care is a rising issue attributable to medical detection methods. It is possible to diagnose up to one-third of cancer diagnoses identified through screening, and 30% of individuals diagnosed with asthma do not have the condition. Testing and overtreatment are also standard; more than 50% of tests were needless before cataract surgery, for example, and were likely to worsen their complications. In particular, in developed nations, antibiotics prescription is still a prevalent issue due to insufficient health staff training. In health management systems, medical AI technologies can help to detect needless diagnostics and procedures. Therefore for large-scale enterprises, we need some hybrid technology that works on controlling health services. Finally, we concluded that the new advances in the intelligence sector had taken place. The devices also dramatically increased the capability and reliability of numerous medical diagnostic techniques. There may be occasions where the issue has been linearized, adding to the reliability of the diagnosis. For nonlinear modelling, different methodologies are being developed. CNN and ANN have been used in numerous applications and have demonstrated that the approaches are considered adaptive to the initialization of the number and forms of neurons and weights. So for excellent precision, the data set is mandatory. This is also not available for well-being since it takes a long time to create such an extensive database.

3 Hybrid Intelligent Medical Diagnosis System Biomedical engineering is a fast-growing field because of the needs and increases in automation. Collaboration between medical professionals and engineers is required to build smart systems for various bio-medicine tasks. In order to detect various diseases, these instruments are used. To help doctors understand diseases’ presence, these tasks serve as clinical decision support systems (CDSS). They, therefore, act as valuable instruments in disease research for physicians. This is especially important given the workload on doctors and the massive prevalence of diseases. The increasing awareness of health has further emphasized the production and use of such systems among people. We use soft computing techniques for disease detection here. This is a classification issue where the aim is to classify the parameters in one of the two groups that indicate whether the disease is present or absent. In order to execute this classification, the device is given several inputs or attributes. Suppose these characteristics are displayed graphically in each class on a graph with a different symbol. There are as many attributes to this graph as the sum of input space. The classification issue’s primary task is to fragment the input space’s distinctive areas so that each class is part of a particular section. The borders of the class are known as judgement boundaries. Based on data or information, every system tries to calculate or predict these constraints. As they have their problems and complications, the classification problems have always merited special mention from the scientific community. One of the curious things about these topics is that while

Hybrid Intelligent System for Medical Diagnosis in Health Care

35

the mechanism and means of classification have been revolutionized, compared to accuracy, the artificial systems seem to be well behind the human classification. A corresponding vocabulary is a sequence corresponding to some particular pattern or attribute’s capacity to distinguish. Machine learning is an exciting environment for researching historical records. In most of the issues, a lot of background data is available. The systems are developed using system-specific training algorithms to learn this data. Learning means removing laws or patterns from historical records. Well-trained programmes can deliver the correct outcomes for the concerns they have been trained for. In comparison, the time and memory constraints would be slightly smaller since the device already has historical data summarized by specific patterns or regulations. Generalization is the tendency of the method to generate the correct outcomes for unknown problems that are identical. This happens by imposing the system’s undefined inputs with extracted patterns or rules. If it is somewhat generalized, a process is considered efficient. In comparison, the time and memory constraints would be slightly smaller since the device already has historical data summarized by specific patterns or regulations. Generalization is the tendency of the method to generate the correct outcomes for unknown problems that are identical. This happens by imposing the system’s undefined inputs with extracted patterns or rules. If it is somewhat generalized, a process is considered efficient. The field that works here is soft computing that comprises the ANN, evolutionary systems and fuzzy inference systems. One of the networks considered to be motivation from the human brain is an artificial neural network [37]. The brain includes an enormous number of corresponding processing components known as neurons. The data is received by each neuron by way of electrical signs, forms the data and further transmits the information to prepare each other neuron. The neural system includes various counterfeit structured neurons that work in an incrusted style. Each neuron gets the total sum of the approaching data and then moves it by an initiation work before contributing it to the following neuron. By and large, we utilize three-layered engineering, which involves a passive input layer, a hidden layer and an output layer. The ANNs structure a decent taking in implies from the critical information (or AI) and sum up the patterns which are found out into the obscure data sources. These systems additionally experience two different degrees of preparing and testing. One of the usually utilized methods for preparing the ANNs is the backpropagation algorithm (BPA). It is a managed learning model where the outputs are known for each information when the training is performed. BPA consists of two different passes, which are the forward pass and the reverse pass. The forward pass is a customary neural network where the data sources are used before assessing the comparing yields for the accompanying. The purposeful publicity of any learning calculation is to locate the worldwide optima in the inquiry space. Another method for adequately addressing the issue incorporates fuzzy surmising frameworks (FIS) [38]. To demonstrate the issue, we need to utilize a fuzzy set hypothesis. According to the fuzzy set hypothesis, each component has a place with some class by a specific degree known as the participation esteem. The information included is taken from the class, which is related somewhat, and there must be a

36

M. Sharma et al.

class of degree that does not have a place with the class. The framework can map the information sources powerfully to deliver yield by the fuzzy rules applications. Fuzzy rules are considered a stimulus compared to the ancient production systems that use the rules to be used on the state to adopt the final state or output. The serious issue that slacks with fuzzy rules are that it works on people’s parameters and rules. The fixed human characteristics are slanted not to be right or dangerous. These frameworks should be physically prepared and tuned, which is not frequently practical for all the issues. This spot a major restriction over the viable utilization of these frameworks in down to earth applications. The constant difference in rules, fixing of framework parameters, and so forth to get the planned framework to proceed according to the excellent yields might be exceptionally troublesome physically. The algorithms that are considered the best for optimizing and searching for the problems are the evolutionary algorithms (EA) [39]. These algorithms are considered nature-inspired. The population is considered the basis on which the evolution of life surrounds, so there is a need to generate the next generations. The fittest one will survive according to Darvin’s theory, and the fitter one will evolve and move to the next generation. The reproduction of the individuals will be more robust who are fitter than the one which is the weaker. The various solutions to the individual’s problem are modelled in the evolutionary algorithms. The new population of individuals are better and optimal compared to the population of the old one. So, the optimal solution to the problem is finding the individual who is considered the best. Different evolutionary operators are motivating by nature and are considered better to create the youngster populace’s ideal populace from the parent’s populace. These calculations are considered powerful to scan for the worldwide minima, which are limited and constrained. This issue can be measured by using these frameworks. If the knowledge size is thought to be immense and the system is considered very complex, the problems are significant. These systems are often considered fragile, and a single mechanism is preferably unable to solve the problems. The method is then implemented to combine the system’s aspects positively, and one of the solutions may be called to eliminate the negative aspects. Harmful elements may be obsolete by the positive component of the scheme. We are concerned with the first form of implementation in Sect. 3.1, i.e. ANFIS. In Sect. 3.2, we address numerous modern methods and new assembly techniques. In evolutionary neural networks, Sect. 3.3 is devoted to using a connectionist approach. We will define the problem-solving technique along with the database in Sect. 3.4. In this chapter, with the assistance of an intellectual device, we explain the different strategies used to diagnose diseases.

3.1 Adaptive Neuro-Fuzzy Inference Systems (ANFIS) Adaptive neuro-fuzzy inference systems (ANFIS), a hybrid framework considered the method of problem-solving, is believed to be the first approach [40]. It is possible to view this method as the integration of ANN with FIS. The fuzzy inference method (FIS), which will be constructed over the neural network design, is also called ANFIS.

Hybrid Intelligent System for Medical Diagnosis in Health Care

37

Fig. 8 ANFIS layered architecture

This will allow us to use the algorithms used in FIS training for a historical database. It will tune FIS and sees it as capable of behaving according to the problem’s demands and providing planned outcomes to issues if it happens to be exposed to them. Mainly, it leads to minimizing errors and achieving excellent problem-solving results. In traditional FIS, automatic tuning of the fuzzy parameters is also an issue that is performed manually. The structure overall is shown in Fig. 8. It is a neural system model with several layers. In both ways, for example, the whole architecture can now be seen as a neural system that has nodes organized in different layers and each layer is relegated to a specific errand to execute. Another FIS will carry out fuzzy activities to measure the yields according to the sources of knowledge. To solve the problem in the above system, as in figure shown (i.e. Fig. 8), it consists of steps. The underlying advance is making an underlying fuzzy model according to the prerequisites of the issue. This may be done physically by the client according to his comprehension of the framework, or consequently, where all the mix of rules might be produced. The main fuzzy model would comprise many participation capacities for each info variable. The quantity of enrolment work assumes an indispensable job in choosing the summing up capacity of the framework. The littler the number of membership functions, the bigger would be the summing up ability. The higher number of part transport capacities makes the issue increasingly confined in nature. It might be noted here that summed up frameworks attempt to discover exceptionally broad principles that fit in the whole arrangement of preparing information. This is opposed to the limited frameworks attempt to outline the standards for some piece of the database. The entire preparing data set might be said as made out of numerous such sets. Speculation is viewed as explicit to the problem as well. All issues cannot be tackled from the most extraordinary level of speculation.

38

M. Sharma et al.

We have to restrict it by using a planning estimate after the model has been constrained. For this, we will use the authentic database of known sources of knowledge and yields. This is a type of supervised method of learning used by ANFIS. Two frequently applied learning algorithms are the back-propagation algorithm and the mixed learning algorithm. The back-propagation algorithm is the same as that described in Sect. 3.1. This attempts to train the algorithm through the gradient descent of the errors in the error space. The test requires providing unknown inputs to the algorithm and separating the data according to the training performed. This stage determines the algorithm’s efficiency.

3.2 Ensemble Approaches The ensemble is carried out as the second hybrid approach or process that can be used for disease detection and diagnosis. A sort of modular neural network (MNN) [40] is the ensemble. The MNN expects to leverage the modularity of the problem by categorizing it into another group of modules. These modules function independently and lead to a complex problem being solved. The old ANN suffers from issues that result in execution dissatisfaction when the planning database is extremely large or the issue is particularly muddled. It hampers their success and also contributes to very long training cycles. MNNs can solve the problem by modularity in the MNNs. The first step to solve an issue is to divide the problem via MNNs into different modules. This splitting takes the same time as the system takes itself while being created. When the issue is given to the unit, it is divided first into the various modules. Ensembles, particularly for such classifying problems, are considered a widely used type of MNNs [41]. In this section, we will only discuss the basic model that is being used for problem-solving. There is also no mention of other approaches and techniques. The first step is to train the machine. All the training data in the system is trained by each module or the ANN independently. Both ANNs must have planned in parallel. Here, each ANN’s output is regarded as the probability vector that denotes each class’s presence or absence. The system states that there are N categories. These modules are researching the issue and working on a response. At that point, they return a probability vector containing probability events for each class. Each party restores the probability vector where each pi indicates the likelihood of class I being present if the classification problem had n classes for which the data might have a location. The case may not be the same because of the system’s weakness but differs if ideally only one out of the considerable number of numbers right now be one, and all others must have −1. According to the principles of the MNNs, the various types of probability vectors are given to the focal integrator. The option of the system’s last yield begins when the integrator gets all the likelihood vectors. The first assignment is to combine the probabilities of the different likely vectors. Henceforth, this represents the last vector of probability that consolidates a large number of modules’ aftereffects. Because of a portion in which the integrator collects the modules’ reactions to the problem to get

Hybrid Intelligent System for Medical Diagnosis in Health Care

39

Fig. 9 Probability-based approach using ensembling

the expected reaction. The integrator selects the class at that stage that gets the most important reaction or has the most serious possibility. The class will be returned to the last class. The notion appeared in Fig. 9. Parallel use of various models or ANNs is the principal aspect of the ensembles. We know that any ANN will not always be able to train itself as per the issue. It can be due to the magnitude of the problem or the data’s size and noise. Also, it might not be prudent to add neurons to the system because they have a detrimental effect on the system’s generalization ability. We, therefore, use several ANNs. Each ANN can now have a different configuration. Differences could be in weight, bias, design, or even the ANN model used in its entirety. As a consequence, various skills and performances are seen. All can be said to specialize in various ways. We must build an integrator to make the most of each module or ANN’s power. Other ANNs eliminate the sub-optimality in the preparation of the ANNs. The entire method, therefore, provides a better outcome for the problem. This makes valuable ensembles tools to solve issues that work badly when single ANNs are used. This notion is shown in Fig. 10.

3.3 Evolutionary Artificial Neural Network Evolutionary neural networks are the last hybrid approach that we use to solve the problem. We train the artificial neural network in this technique using the genetic algorithm. ANN’s training uses BPA, but many limitations are overcome by using

40

M. Sharma et al.

Fig. 10 Ensemble using a polling approach

GA. The ANN might suffer from the local minima problem. The algorithm design was defined initially, which could not be optimized according to the problem [42, 43]. The methods used by evolutionary ANN develop and train the ANN using the genetic algorithm’s optimization property. The ANN approach optimizes the methodology in which mistake or output can play an important role. The weights, prejudices and the framework are the modifiable parameters. A fully associated model in which each neuron of each layer has a complete relationship to the neurons of the associated layers is the ANN we consider to solve the problems shown in Fig. 11. This arrangement results in an excessive demand for calculation that increases the overall time for preparation. Neuron subtraction may not be that important as the network may not train fewer neurons. If the scheme fails to plan, the general methodology at that point is the expansion of a neuron. Be that as it may, the enormous increase in size may expand a solitary neuron that will increase the dimension by an enormous amount. This will make it difficult to prepare the device for the training calculation because of the equivalent. Henceforth, we use the idea of a connectionist approach as a general multifaceted nature of the problem that will try to hinder or prevent such connections by limiting the calculation time. This makes the forecasts in the blink of an eye and much more; since the similarities are confined, preparation will eventually show itself. We will usually use GA to establish the connectionist framework during this downside, despite settling ANN loads. Nevertheless, the most intense range of neurons must be highlighted at the start. The first portion of the body consists of a zero or one that depends on an interaction between the input layer and the hidden layer’s neurons. One reflects the existence

Hybrid Intelligent System for Medical Diagnosis in Health Care

41

Fig. 11 ANN architecture

of the association, and a zero suggests the lack of association. Amid a hidden nerve layer cell and an output nerve layer cell, the following section also shows weight. The bits have an indicative constant. The opposite half requires the basic weight values between the various relations of the body. The last half consists of the weights between the input and hidden layer neurons, hence the second segment between the hidden and output layer neurons. These squares test the basic values of associations that hold whether or not the relationship exists. The corresponding weight value is also ignored if the physical relationship is not there. It includes the values of the biases in the last half. The most important factor to use the GA algorithm is the genetic operators. This algorithm helps in the production of the individuals to achieve them to greater heights. The selection is considered the genetic operator, and it deals with selecting the individuals to go to a higher generation. The theory of selecting the fittest individual and moving to the next generation is considered Darwin’s theory, which says only the fittest individual will survive and move to the next generation. From the population pool, only the individual who is the fittest will survive. The operator used is a crossover, and the function of this operator is to mix the parent’s solutions and form a child chromosome. The modification of this operator is required to fulfil the requirement of the problem. With the help of mutation, we can add new characteristics to the individual. If the characteristics added are good, then the algorithm will work, and the new off springs will be good. However, if the added characteristics are not good, then the individual will be removed by the concept of survival of the fittest. The fitness function is a function that can measure how good or optimal the solution is. The inhabitants are improved by the GA in its run when the fitness function

42

M. Sharma et al.

calculates the optimality. Every individual is considered as an ANN which predicts the presence of disease.

3.4 Application of Hybrid Intelligent System in Health Care The hybrid intelligent system must solve problems like diagnosing and detecting diseases in the preliminary stages. Many diseases can be detected using these systems, which have various applications in various diseases. We need a data set from the UCI machine learning repository [44] for implementation. The UCI machine learning data is obtained from the repository of machine learning. Depending on the device, it is sent for training after the elicitation of data using the various training algorithms. The system is fed with the new data selected from the sample data set. The output of the system is collected, and the comparison is made with the standard output. After the data is trained, it is sent for testing. The systems are tested for various parameters like performance, efficiency, ability and evaluation [45]. The general methodology is summarized by Fig. 12. Fig. 12 General working methodology

Hybrid Intelligent System for Medical Diagnosis in Health Care

43

Hence, it is based on the optimization and automation of IT processes and the technical environment of interrelated objects. This environment aims to develop the current procedures for providing advanced medical care and open up new medical possibilities. The automated system’s fundamental principle includes efficiency, quality improvement, expansion of the patient’s capabilities, accessibility for everyone, etc.

4 Need of Hybrid Intelligent System in Medical Diagnosis The hybrid intelligent machine becomes increasingly difficult to do what people do, but more effectively, more efficiently and at less cost. In several real-world applications, many intelligent technologies need to be implemented. Each technology compensates for the other technology’s weak areas. A smart hybrid system combines two smart technologies. The combination of a neural network with a fugitive technique, for example, leads to a hybrid neurofugal technique. The combination of probability reasoning and neural networks thus forms the foundations or essential components of soft computing. This soft machine is considered a developmental viewpoint to construct hybrid intelligent systems in an unresolved and unclear sense, reason and understand. The scope of social insurance for simulated intelligence and mechanical autonomy is immense. Like our consistent lives, computer-based information and mechanical technology are increasingly a part of our eco-framework for medical services. We have featured eight different ways that exhibit how this change is right now in progress, as shown in Fig. 13. (i)

(ii)

Keeping Well One of the biggest possible benefits of the hybrid intelligent system is to assist citizens with staying solid because they do not need to bother with a professional, or if nothing else, not as often. As of now, the use of a half-breed-wise system in shopper well-being applications benefits individuals. Applications and applications for creativity promote more positive action in individuals and help proactively manage a sound way of life. This puts consumers in control of health and prosperity. It also improves human services professionals’ willingness to consider the common examples and necessities of the people they care about. They can provide better feedback, direction and backing for remaining sound with that understanding. Early Detection To recognize disease all the more precisely and in the beginning times, half and half canny framework is probably the best framework to distinguish it in the beginning periods. As indicated by American Malignant growth Study (ACS), a great extent of X-Beams yields fake results encouraging 1 out of 2 sound ladies to be told that they have the disease. These frameworks’ proper use gives quicker outcomes, best review and exactness esteems planning to accomplish around 99% precision, lessening the requirements’

44

M. Sharma et al.

Fig. 13 Hybrid intelligent system

(iii)

(iv)

superfluous assessments. The spread of shopper wearables and other clinical gadgets joined with the mixture smart framework is likewise applied to regulate beginning period coronary illness, empowering specialists and different parental figures to more readily screen and distinguish conceivably perilous scenes at prior, increasingly treatable stages. Diagnosis To unlock vast volumes of health information and capacity diagnosis, IBM’s Watson for health helps service institutions implement psychological feature technologies. Watson can scan and archive more medical evidence infinitely faster than every human—each medical journal, symptom, and case study of care and reaction worldwide. Google’s DeepMind well-being aims to unravel real-world concentration challenges in collaboration with doctors, academics and patients. The technology comprises machine learning and systems neurobiology to transform the general learning algorithms into neural networks that imitate the human brain. Decision-Making Upgrading treatment involves aligning vast patient information with appropriate and timely decisions, and therapeutic decision-making and behaviour and prioritizing body activities can be assisted by predictive analytics. Another area where the hybrid intelligent system starts to take shape in health care is identifying victimization trends to spot patients at risk of acquiring a disease or watching it deteriorate due to modus vivendi, occupational, genomic or alternative causes.

Hybrid Intelligent System for Medical Diagnosis in Health Care

(v)

(vi)

(vii)

(viii)

45

Treatment In addition to scanning health records to aid suppliers in establishing inveterately unwell persons, the United Nations organization is now at risk of AN adverse episode. A hybrid intelligent device would make it possible for physicians to take a holistic approach to illness treatment, help organize clinical arrangements and make it easier for patients to handle their long-run therapy services more efficiently. Robots have been used in medicine for over 30 years. They range from basic laboratory robots to overly sophisticated surgical robots that can support the operative surgeon with a personality or conduct procedures independently. In hospitals and laboratories, they are used for routine activities, recovery, physiotherapy and long-term care for surgery. End of Life Care Firstly, lives have become longer in comparison with previous generations. Second, death is different and slower as the end of life is reached due to varying heart failure, dementia and osteoporosis. Also, isolation has become a part of life. Robots will theoretically change the area of life care, enabling patients to remain free for a prolonged period, thus reducing hospitalization and care home requirements. In conjunction with advances in humanoid design, the hybrid intelligent system has allowed robots to hold ‘conversations’ and numerous social interactions so that ageing minds remain sharp. Research The distance between an exploration laboratory and an actual patient might be long and demanding. In the context of the American state clinical strength investigation affiliation, it takes a mean of twelve years for a medication to reach the patient from the examination research facility. Exclusively 5 out of 5000 special test of the medication are assembled for human testing, and only one of those five is ever endorsed for human utilization. By and large, it will esteem an enterprise US $359 million to build up a substitution tranquillize from the exploration research facility to the patient. Medication investigation and revelation are among the more current applications for the cross-breed astute framework in medicinal services. By managing the most up to date propels inside the half and half insightful frameworks to shape the medication disclosure and medication repurposing forms, there is the possibility to impressively cut each the time and incentive to plug for fresh out of the plastic new medication. Training

The training is done via realistic simulations using hybrid intelligent systems that simple computer algorithms cannot do. The natural language is the advancement and is considered the ability to draw on a big database so that the decisions can be taken as a challenge that humans cannot. The previous responses can be used as a training programme to make them learn. The training can also be done anywhere via the hybrid intelligent systems that can be used on smartphones.

46

M. Sharma et al.

Table 1 Intelligent technologies for the solution to healthcare problems S. No. Category Intelligent techniques can solve many healthcare problems such as skin diseases, different types of cancers related problems, Sepsis attack, Alzheimer’s, Parkinson’s, chronic kidney diseases and many more

Some approaches Feed-forward neural network Modular neural network Quantitative approach Back-propagation neural network\ Fuzzy expert system Adaptive neuro-fuzzy inference systems (ANFIS) Evolutionary artificial neural network Recurrent neural network Ensemble approaches

Early disease identification is one of the biggest obstacles faced by healthcare providers. Therefore, they are a wide variety of illnesses such as cancer, heart attacks, sepsis attack, Alzheimer’s, Parkinson’s and COVID-19, among others. The precision of many conventional applications is not adequate to disclose this type of disease. This chapter seeks to implement an intelligent model which combines two or more technologies to tackle this challenge. Below is the name of many intelligent techniques that can be applied for the solution of many healthcare problems which is shown in Table 1.

5 Conclusion There are the tremendous scope and opportunity for developments made in the hybrid intelligence system to revolutionize healthcare practices. This chapter’s essential purpose was to make one aware of the implications of recent hybrid technology and the ongoing research and its findings on hybrid computational intelligence applications in healthcare practices. This chapter will offer readers a broad and consistent perspective on the development and usage of this area. The chapter offers a thorough insight into how hybrid intelligent systems are much better than the conventional method. First, the chapter discusses the need for an intelligence structure followed by a comprehensive hybrid intelligence description and then a comparative analysis between the two. A thorough insight into health care is also presented, which also illustrates the progress that health care has made by introducing intelligence system techniques. The need for both the intelligence system and the healthcare sector’s hybrid intelligence system was then met. With such a fast healthcare transition rate, healthcare institutions increasingly adapt to the new technology to control and fulfil patient demands. The artificial intelligence (AI) field provides a fantastic opportunity to learn from the past and make better future decisions, as seen above. The genetic algorithm, neural networks, fuzzy

Hybrid Intelligent System for Medical Diagnosis in Health Care

47

logic, machine learning and various basic computational intelligence techniques are hybrid to each other to construct the hybrid method. The continued emphasis and attention on quality, cost and treatment would ensure that developments are made in various AI innovations to increase the quality of the services delivered across different healthcare systems.

References 1. Meesad, P., & Yen, G. G. (2003). Combined numerical and linguistic knowledge representation and its application to medical diagnosis. IEEE Transactions on Systems, Man and Cybernetics, Part A: Systems and Humans, 33(2), 206–222. https://doi.org/10.1109/TSMCA.2003.811290 2. Pavlopoulos, S. A., & Delopoulos, A. (1999). Designing and implementing the transition to a fully digital hospital. IEEE Transactions on Information Technology in Biomedicine, 3(1), 6–19. https://doi.org/10.1109/4233.748971 3. Chabat, F., Hansell, D. M., & Yang, G. (2000). Computerized decision support in medical imaging. IEEE Engineering in Medicine and Biology Magazine, 19(5), 89–96. https://doi.org/ 10.1109/51.870235 4. Tsipouras, M. G., Voglis, C., & Fotiadis, D. I. (2007). A framework for fuzzy expert system creation—Application to cardiovascular diseases. IEEE Transactions on Biomedical Engineering, 54(11), 2089–2105. https://doi.org/10.1109/tbme.2007.893500 5. Kwiatkowska, M., Atkins, M. S., Ayas, N. T., & Ryan, C. F. (2007). Knowledge-based data analysis: First step toward the creation of clinical prediction rules using a new typicality measure. IEEE Transactions on Information Technology in Biomedicine, 11(6), 651–660. https://doi. org/10.1109/titb.2006.889693 6. Isola, R., Carvalho, R., & Tripathy, A. K. (2012). Knowledge discovery in medical systems using differential diagnosis, LAMSTAR, and k-NN. IEEE Transactions on Information Technology in Biomedicine, 16(6), 1287–1295. https://doi.org/10.1109/TITB.2012.2215044 7. Çomak, E., Polat, K., Güne¸s, S., & Arslan, A. (2007). A new medical decision-making system: Least square support vector machine (LSSVM) with fuzzy weighting pre-processing. Expert Systems with Applications, 32(2), 409–414. https://doi.org/10.1016/j.eswa.2005.12.001 8. Luukka, P. (2011). Feature selection using fuzzy entropy measures with similarity classifier. Expert Systems with Applications, 38(4), 4600–4607. https://doi.org/10.1016/j.eswa.2010. 09.133 9. Kahramanli, H., & Allahverdi, N. (2009). Extracting rules for classification problems: AIS based approach. Expert Systems with Applications, 36(7), 10494–10502. https://doi.org/10. 1016/j.eswa.2009.01.029 10. Sekar, B. D., Dong, M.-C., Shi, J., & Hu, X. Y. (2012). Fused hierarchical neural networks for cardiovascular disease diagnosis. IEEE Sensors Journal, 12(3), 644–650. https://doi.org/ 10.1109/JSEN.2011.2129506 11. Markowska-Kaczmar, U., & Matkowski R. (2006). Experimental study of evolutionary based method of rule extraction from neural networks in medical data. In P. Perner (Eds.), Advances in data mining. Applications in medicine, web mining, marketing, image and signal mining. ICDM 2006. Lecture notes in computer science (Vol. 4065, pp. 76–90). Springer. https://doi. org/10.1007/11790853_7. 12. Economou, G. P. K., Goumas, P. D., & Spiropoulos, K. (1996). A novel medical decision support system. Computing and Control Engineering Journal, 7(4), 177–183. https://doi.org/ 10.1049/cce:19960404 13. Kovalerchuk, B., Vityaev, E., & Ruiz, J. F. (2000). Consistent knowledge discovery in medical diagnosis. IEEE Engineering in Medicine and Biology Magazine, 19(4), 26–37. https://doi. org/10.1109/51.853479

48

M. Sharma et al.

14. Kinney, W. C. (2003). Web-based clinical decision support system for triage of vestibular patients. Otolaryngology-Head and Neck Surgery, 128(1), 48–53. https://doi.org/10.1067/mhn. 2003.33 15. Huang, K., Yang, H., King, I., & Lyu, M. R. (2006). Maximizing sensitivity in medical diagnosis using a biased minimax probability machine. IEEE Transactions on Biomedical Engineering, 53(5), 821–831. https://doi.org/10.1109/TBME.2006.872819 16. Zadeh, L. A. (1998). Roles of soft computing and fuzzy logic in the conception, design and deployment of information/intelligent systems. In O. Kaynak, L. A. Zadeh, B. Türk¸sen, & I. J. Rudas (Eds.) Computational intelligence: Soft computing and fuzzy-neuro integration with applications. NATO ASI Series (Series F: Computer and Systems Sciences) (Vol. 162). Springer. https://doi.org/10.1007/978-3-642-58930-0_1. 17. Kandel, A., & Langholz, G. (Eds.) (1992). Hybrid architectures for intelligent systems. CRC Press. ISBN-9780849342295. 18. Grosan, C., & Abraham, A. (2011). Hybrid intelligent systems. In: Intelligent systems. Intelligent systems reference library (Vol. 17). Springer. https://doi.org/10.1007/978-3-642-210044_17. 19. Jouffe, L. (1998). Fuzzy inference system learning by reinforcement methods. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 28(3), 338–355. https://doi.org/10.1109/5326.704563. 20. Procyk, T. J., & Mamdani, E. H. (1979). A linguistic self organising process controller, automatica. Journal of IFAC, 15(1), 15–30. https://doi.org/10.1016/0005-1098(79)90084-0. 21. Wang, L. X. (1994). Adaptive fuzzy systems and control. Prentice Hall Inc. ISBN 978-0-13099631-2. 22. Wang, L. X., & Mendel, J. M. (1992). Backpropagation fuzzy system as nonlinear dynamic system identifiers. In Proceedings of the First IEEE International Conference on Fuzzy Systems, San Diego, USA (pp. 1409–1418). https://doi.org/10.1109/FUZZY.1992.258711. 23. Sugeno, M., & Tanaka, K. (1991). Successive identification of a fuzzy model and its applications to prediction of a complex system. Fuzzy Sets Systems, 42(3), 315–334. https://doi.org/10.1016/ 0165-0114(91)90110-C. 24. Abraham, A., & Nath, B. (2000). Evolutionary design of fuzzy control systems—A hybrid approach. In J. L. Wang (Ed.), The Sixth International Conference on Control, Automation, Robotics and Vision (ICARCV 2000), CD-ROM Proceeding. ISBN 9810434456. 25. Pedrycz, W. (Ed.). (1997). Fuzzy evolutionary computation. USA: Kluwer Academic Publishers. 26. Sanchez, E., Shibata, T., & Zadeh, L. A. (Eds.). (1997). Genetic algorithms and fuzzy logic systems: Soft computing perspectives. Singapore: World Scientific Publishing Company. 27. Cord´on, O., Herrera, F., Hoffmann, F., & Magdalena, L. (2001). Genetic fuzzy systems: evolutionary tuning and learning of fuzzy knowledge bases. World Scientific Publishing Company. https://doi.org/10.1142/4177. 28. Abraham, A. (2003). Business intelligence from web usage mining. Journal of Information and Knowledge Management (JIKM), 2(4), 1–15. https://doi.org/10.1142/S0219649203000565 29. Abraham, A. (2003). i-Miner: A web usage mining framework using hierarchical intelligent systems. In The 12th IEEE International Conference on Fuzzy Systems FUZZ—IEEE’03 (Vol. 2, pp. 1129–1134). IEEE Press. https://doi.org/10.1109/FUZZ.2003.1206590 30. Bhattacharya, A., Abraham, A., Grosan, C., Vasant, P., & Han, S. (2006). Meta-learning evolutionary artificial neural network for selecting flexible manufacturing systems. In J. Wang, Z. Yi, J. M. Zurada, B. L. Lu, & H. Yin (Eds.), Advances in neural networks—ISNN 2006. Lecture notes in computer science (Vol. 3973). Springer. https://doi.org/10.1007/11760191_130 31. Millman, M. (1993). Access to health care in America (1993). The National Academies Press, US National Academies of Science, Engineering and Medicine. ISBN-10: 0-309-04742-0. 32. Kjoelen, A., Thompson, M. J., Umbaugh, S. E., Moss, R. H., & Stoecker, W. V. (1995). Performance of AI methods in detecting melanoma. Engineering in Medicine and Biology, 14(4), 411–416. https://doi.org/10.1109/51.395323

Hybrid Intelligent System for Medical Diagnosis in Health Care

49

33. Schölkopf, B., Burges, C., & Vapnik, V. (1995). Extracting support data for a given task. In U. M. Fayyad, & R. Uthurusamy (Eds.), First International Conference on Knowledge Discovery and Data Mining, Proceedings. AAA I Press. 34. Vapnik, V. N., Golowich, S. E., & Smola, A. (1998). Support vector method for function approximation, regression and signal processing. In Advances in neural information processing systems (Vol. 9, pp. 281–287). MIT Press. 35. Singh, H., Meyer, A. N. D., & Thomas, E. J. (2014). The frequency of diagnostic errors in outpatient care: Estimations from three large observational studies involving US adult populations. BMJ Quality and Safety, 23(9), 727–731. https://doi.org/10.1136/bmjqs-2013-002627 36. Ridley, E. L. SIIM: AI poised to enhance all aspects of radiology. 2017. Available at www.aun tminnie.com/index.aspx?sec=sup&sub=aic&pag=dis&ItemID=117495! Accessed 21 March 2018. 37. Ciampi, A., & Zhang, F. (2002). A new approach to training back-propagation artificial neural networks: Empirical evaluation on ten data sets from clinical studies. Statistics in Medicine, 21, 1309–1330. https://doi.org/10.1002/sim.1107 38. Xiaoguang, C., & Lilly, J. H. (2004). Evolutionary design of a fuzzy classifier from data. IEEE Transactions on Systems, 34(4), 1894–1906. https://doi.org/10.1109/TSMCB.2004.831160. 39. Srinivasaa, K. G., Venugopala, K. R., & Patnaikb, L. M. (2007). A self-adaptive migration model genetic algorithm for data mining applications. Information Sciences, 4295–4313. https://doi. org/10.1016/j.ins.2007.05.008. 40. Shadabi, F., Sharma, D., & Cox, R. (2006). Learning from ensembles: Using artificial neural network ensemble for medical out-comes prediction. In IEEE Conference: Innovations in Information Technology. https://doi.org/10.1109/INNOVATIONS.2006.301896. 41. Kala, R., Shukla, A., & Tiwari, R. (2009). Fuzzy neuro systems for machine learning for large data sets. In Proceedings of the IEEE International Advance Computing Conference (pp. 541–545). https://doi.org/10.1109/IADCC.2009.4809069. 42. Augusteijn, M. F., & Harrington, T. P. (2004). Evolving transfer functions for artificial neural networks. Neural Computing and Applications, 13(1), 38–46. https://doi.org/10.1007/s00521003-0393-9 43. He, S., Wu, Q. H., & Saunders, J. R. (2006). A group search optimizer for neural network training. In Proceedings of Computational Science and Its Applications (pp. 934–943). ICCSA. https://doi.org/10.1007/11751595_98. 44. Sigillito, V. G., Wing, S. P., Hutton, L. V., & Baker, K. B. (1989). Classification of radar returns from the ionosphere using neural networks. Johns Hopkins APL Technical Digest, 10, 262–266. 45. Shukla, A., Tiwari, R., Meena, H., & Kala, R. (2009). Speaker Identification using wavelet analysis and modular neural networks. Conference Proceedings of National Symposium on Acoustics, 36(1), 14–19.

Remote Patient Monitoring Using IoT, Cloud Computing and AI M. V. V. Prasad Kantipudi, C. John Moses, Rajanikanth Aluvalu, and Sandeep Kumar

Abstract Innovative wireless communication technologies using body sensors and the advent of Internet of things (IoT) are used to introduce many modern healthcare schemes to provide personalized health management and prevent some acute diseases. Monitoring and caring for patients remotely using modern techniques such as IoT, cloud computing (CC), and artificial intelligence (AI) are evolving proposals on healthcare innovations. Over the past decade, taking care of patients through remote access is widely suggested by different authors to monitor patients affected by different illnesses like heart disease, neurological sickness, blood pressure, body temperature, chronic disease, diabetes, and obesity. Further, remote monitoring is used to care for post-operative patients and aged patients using smart sensors and intelligent decision-making technologies. CC is a sophisticated technology comprised of remote servers which act as a gateway of remote access by connecting intelligent sensors and intelligent devices with the concept of IoT. AI is a genius technique of making decisions using deep learning (DL) methodology with the cloud dataset. This chapter illustrates the concept of cloud and AI-based IoT for remote health caring. It also discusses different decision-making systems using AI and the principle of operation of several cloud infrastructures used to access secured medical records. By learning this chapter, the readers and young researchers will understand the principle and challenges of IoT, CC, and AI in various healthcare schemes to identify the suitable architecture of the cloud and AI for different disease diagnosis and patient monitoring. Keywords Remote access · Smart sensors · Intelligent decision · Remote servers · Blockchain

M. V. V. P. Kantipudi (B) · C. J. Moses · S. Kumar Sreyas Institute of Engineering and Technology, Hyderabad, India R. Aluvalu Vardhaman College of Engineering, Hyderabad, India © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 A. Kumar Bhoi et al. (eds.), Hybrid Artificial Intelligence and IoT in Healthcare, Intelligent Systems Reference Library 209, https://doi.org/10.1007/978-981-16-2972-3_3

51

52

M. V. V. P. Kantipudi et al.

1 Cloud-Oriented IoT Using AI 1.1 Introduction to Internet of Things Day by day, IoT plays a vital role in human life and it has the power to change our lives significantly. IoT is extending the strength of the Internet beyond human expectations. IoT refers to the physical devices connected to the internet and all collecting and sharing data. IoT is connecting the entire smart device through the Internet to communicate all the devices correctly. In other words, we can say IoT is made up of smart sensors, devices and wearable connected. Nowadays, the arrival of supercomputer chips and the ubiquity of wireless networks make it possible to turn anything into a part of the IoT. It is one of the fast-growing and widely used technologies in our fast-track carrier. IoT grows day by day because more than 50 billion devices will be connected to the Internet by 2020 and by 2025 IoT market is expected to reach 6.2 trillion dollars. Figure 1 shows the growth of IoT in terms of the number of connected devices. IoT consists of two simple words. The first one is the Internet, and the second one is the thing. All among these two words. Anything is connected to the Internet of things and everything in smartphones, cars, laptops, microwaves, and heart monitor implants. All of these are taken as an application of IoT. It describes a pure interconnected world. In worldwide, every device is developed with smart capabilities to connect, interact, and exchange data. It is a world with technology more straightforward, simple, and better. Top ten Internet of things (IoT) trends that will rule in 2020, Voice Assistants: In a smart house, IoT allows all smart devices to communicate or travel from area to area.

Fig. 1 Growth of IoT in terms of number of connected devices

Remote Patient Monitoring Using IoT, Cloud Computing and AI

53

Wearables Are Finally Diversifying: A smart wearable device provides health and fitness status to doctors for instant access to saving our lives in daily life. Smart Mirror on the Wall: A smart mirror can show the day-by-day updates, i.e., weather, news, and work schedule. Smart Home Market: In a smart home, IoT-based smart devices can reduce power, time, energy, and costs. Do not Forget About Self-Driving cars: Top automotive companies are ambitious on IOT-based self-driving cars. Behold, The Flying Taxi: The flying car or taxi will start for traveling in UAE, and it will be transformed sooner in the whole world. Edge Computing: The edge computing technology will improve response times, and it will save bandwidth. Flexible Displays: The flexible displays technology will play a significant role in the decade. 5G: Obscuring Reality: 5G technology designed with cellular network concept, and this IoT application will make a big difference. Sensor Innovation: Billions of smart sensors are interconnected through smart sensors to collect the surrounding environment’s data. The functional structure of IoT is shown in Fig. 2. As shown in Fig. 2, the plant is any physical system or environment that can interact with the IoT system. IoT consists of a group of devices connected from the leaves in the network. A node consists of sensors/actuators, memory, processor, and every node is connected through an interface. A node may run the protocol of IoT. Hubs produce the first-level connection between the network and nodes. Hubs are usually run IP and fog processor operates on a set of local nodes and hubs. Further,

Fig. 2 Structure of IoT

54

M. V. V. P. Kantipudi et al.

cloud servers present computational services of the IoT structure. Databases are used to store general information and the results of the computation. The cloud can interact between the users and nodes [1, 2].

1.2 Cloud Computing (CC) Cloud computing is a pool of configurable resources, i.e., storing, networks, applications, and organizations, that can be immediately conveyed with unimportant organization endeavors. Distributed computing would be a profitable market if all around used [30]. Cloud computing is utilized to get to a bunch of registering assets. The web keeps it up. Utilizing these assets, clients can get to foundations, figuring force, applications, and administrations on request. It empowers clients to get to their data, wide range of applications, and information anyplace dependent on their requirements. Moreover, cloud computing is implied by the straightforward transformation of equivalent and conveyed systems used for sharing, assurance, and complete of topographically disseminated ‘self-administering resources intensely at runtime depending upon their capacity, execution, and customers’ nature of administration necessities [3]. These days advancement of cloud computing is extraordinary in the IT region. It ensures the limit of a broad scope of data on the web safely. Currently, cloud computing research is the emerging area for the clinical field, research field, huge tech and medications, and it is a purposeful anecdote for the web [4]. In the digital world, smart gadgets are used to get all social orders through the web. So distributed computing plays a significant role in storing and save this real-time data for various applications, i.e., WhatsApp, Facebook, Twitter, and Amazon. It has developed as a popular answer for a gracefully modest and straightforward way to deal with externalized IT assets. Figure 3 shows an essential distributed computing climate [5].

1.3 Artificial Intelligence (AI) Artificial intelligence (AI) insight the machines can think like human beings and take their own decisions based on their surroundings. It is also called machine intelligence. PCs are also ready to perform errands, such as visual discernment, discourse acknowledgment, dynamic, and interpretation between dialects, ordinarily requiring human knowledge [6]. In programming designing, plans to make contraptions that can re-enact the human capacity to think, see, profoundly customized clients encounter and have the alternative to tackle different kinds of issues. This limit will allow cycles to end up being snappier [7]. With mechanical progressions in AI and its potential, we see

Remote Patient Monitoring Using IoT, Cloud Computing and AI

55

Fig. 3 Cloud computing environment

its extending fuse in whole regions of society, for instance, prescription, public organizations, and the military. Figure 4 [8] shows the eventual fate of AI. Artificial Intelligence’s Trusts: AI standards and scenarios, from education to entertainment [9] as dimensions one may identify with as follows: Job effect: AI makes our life easy, and it will reduce the workload by coping with our work schedules, robotic mechanization of tasks. Education: AI enhances the learning of understudies, for example, through programmed mentoring or reviewing.

Fig. 4 Future of AI [8]

56

M. V. V. P. Kantipudi et al.

Transportation: New modes of transport, such as self-driving cars, flying cars, are allowed by AI. Health care: AI improves human beings’ well-being and prosperity, e.g., by helping to find, reveal medications, or motivate personalized treatment. Creation of decisions: AI or master structures allow us to decide on better choices, e.g., when to hold a meeting or business chiefs’ case-based thinking. Entertainment: AI gives us pleasure by entertainment, e.g., but in video games, more astute adversaries. Singularity: A possible particularity would bring positive social benefits, e.g., interminability.

1.4 Deep Learning Architecture The profound learning concept resolves the real-time issues significantly, and it will provide the best endeavors to human life for a long time. Deep learning provides in-depth information in all the applications, i.e., science, management, government, and business. It has proven better outcomes as compared to other AI techniques. Deep learning has delivered encouraging outperforms in standard language considering, i.e., impression, question noting, language interpretation, etc. The quantity of plans and calculations that are used in profound learning is called profound learning design. The most mainstream profound learning designs are recurrent neural network (RNNs), long short-term memory (LSTM)/gated recurrent unit (GRU), convolutional neural network (CNN), deep belief network (DBN), and deep stacking networks (DSNs)—and afterward investigates open-source programming choices for profound learning. These structures are applied in various circumstances; however, the accompanying table records part of their typical applications [10]. Recurrent neural organization (RNN) The RNN is one of the essential organization plans from which other profound learning structures are planned. An intermittent organization may have associations that contribute to prior layers (or into a comparable layer). This info licenses RNNs to record past information sources and model issues true to form. RNNs contain a rich game plan of engineering. It is chiefly utilized for discourse acknowledgment and penmanship acknowledgment. LSTM/GRU organizations Hochreiter and Schimdhuber proposed the LSTM concept in the year 1997, and later on, this technology came with RNN for various applications. The LSTM is a neuralbased structure to utilize less memory during execution. We are using this concept in our daily life for smart devices and later on in 2014, improvement done in the existing LSTM. For specific applications, the GRU has execution like the LSTM, yet being less troublesome strategies, less loads, and faster execution. It might be

Remote Patient Monitoring Using IoT, Cloud Computing and AI

57

utilized for picture inscribing, signal acknowledgment, and characteristic language text pressure. Convolutional neural organization (CNN) A CNN is a visual cortex usually handled a combination of multiple layers of neural that and CNN plays a significant role in engineering to handling various applications. The LeNet CNN engineering is involved a few layers that execute highlight extraction and afterward order. These layers are the combination of convolutions, filters, pooling, an associated order layer and these layers cleared the path for various new employments of profound learning neural associations. Despite picture preparation, the CNN has been viably applied to video affirmation and various assignments inside characteristic language handling. It is mainly utilized for picture acknowledgment, video investigation, and standard language preparation. Deep conviction organization (DCN) The DBN is standard organization engineering yet fuses a novel preparing calculation. The DBN is including many concealed layer which is related to restricted Boltzmann machine (RBM). The significant utilization of DBN is disappointment forecast, characteristic language understanding, data recovery, and picture acknowledgment. Deep stacking networks (DSN) The final design is the DSN, furthermore called a profound arched organization. A DSN is not precisely equivalent to conventional profound learning structures, and it is a profound course of action of individual organizations. This design resolves one of the issues with profound learning: the multifaceted nature of preparing. Each layer in a profound learning design dramatically expands the multifaceted nature of preparing. It is utilized for persistent discourse acknowledgment and data recovery. Innovative wireless communication technologies using body sensors and IoT are used to introduce many modern healthcare schemes to provide personalized health management and prevent some acute diseases. Monitoring and caring for patients remotely using modern techniques such as IoT, CC, and AI are evolving proposals on healthcare innovations. Over the past decade, taking care of patients through remote access is widely suggested by different authors to monitor patients affected by different illnesses like heart disease, neurological sickness, blood pressure, body temperature, chronic disease, diabetes, and obesity. Further, remote monitoring is used to care for post-operative patients and aged patients using smart sensors and intelligent decision-making technologies. CC is a sophisticated technology comprised of remote servers which act as a gateway of remote access by connecting intelligent sensors and intelligent devices with the concept of IoT. AI is a genius technique of making decisions using DL methodology with datasets available in the cloud. This chapter illustrates the concept of cloud and AI-based IoT for small health caring. It also discusses different decision-making systems using AI and the principle of operation of several cloud infrastructures used to access secured medical records. By learning this chapter, the readers and young researchers will understand the principle and challenges of IoT, CC, and AI in various healthcare schemes to

58

M. V. V. P. Kantipudi et al.

identify the suitable architecture of the cloud and AI for different disease diagnosis and patient monitoring.

2 Wireless Body Networks (WBN) 2.1 Overview The improving usage of a wireless medium of communication and the reduction of electrical invasive and non-invasive devices has motivated wireless body networks (WBNs). The WBN is used to monitor any patients continuously without constraint on the patient’s daily activities. Many technologies have developed to support WBNs applications like biofeedback, remote monitoring and assisted living to meet their high-quality services. This section explains the purpose of different body networks and the supporting sensors and other applications of WBNs.

2.2 Architecture and Applications WBN comprises different genetic sensors, and these sensors are fixed in various parts of the human body. These kinds of setups measure various modifications in human vital signs and determine humans’ emotions like happiness, stress, and fear. The WBN device can communicate using coordinator nodes, typically a low power, consume and have high processing capacities. The WBN is responsible for sending the user’s biological signals to the caretakers to take the required actions. Figure 5 shows the commonly used architecture of WBN communication [11]. Communication between the different levels of WBN is accomplished by using different technologies like Wi-Fi, ZigBee, IEEE 802.15.6, and Bluetooth [12]. InterWBN communications can also be developed using master node and other personal devices like home service robots and notebooks. WBN is widely used for medical and non-medical applications, as shown in Fig. 6 [13]. As shown in Fig. 5 [12], the WBN can diagnose different human diseases and monitor different user features, for example, temperature, pulse, and blood pressure. When the anomalous condition is identified, the wireless sensors’ data can be transferred to the smartphone. This data is directly transferred to doctors or any other help desk for taking the necessary actions. The Internet or a cellular network may be utilized to deliver the data from gateway to destination, as shown in Fig. 5. Further, WBNs are used to take pre-emptive health measures and remote monitoring.

Remote Patient Monitoring Using IoT, Cloud Computing and AI

59

Fig. 5 Four-level communication of wireless body networks

Fig. 6 WBN applications

2.3 Hybrid Sensor-Based Healthcare Systems In the past decade, several enterprises created and distributed to the marketplace various wearable goods in types of bracelets, smartwatches, earphones, and fitness bands for inconspicuous. The health information gathered by these smart gadgets

60

M. V. V. P. Kantipudi et al.

endures irregularities and reliabilities that influence their effectiveness and fitness for medical purposes. Wearable sensors used for health checking techniques may be incorporated into clothes or precisely connected to the human body. The wearable sensors are competent for assessing biological factors such as electromyogram (EMG), electrocardiogram (ECG), body temperature, heart rate (HR), blood pressure (BP) [15–18]. Wearable sensors can be categorized into two kinds such as non-invasive and invasive. Non-invasive wearable sensors are utilizing for non-stop monitoring, and these sensors do not require any physical support with the device, i.e., smart band, watches, and incorporated into textile. Invasive wearable sensors are categorized further as slightly invasive, and these types of sensors need clinical involvement to spot them within the human body [18]. Typically, a WBN node comprises single or extra sensors connected, embedded microprocessor, limited computational capability and memory, power unit, and transceiver unit. These units permit every node to converse with the system. Interaction among the intersections is central—it may be an interacting platform of committed servers or secluded (cloud) servers. This network organization relates to the core of the IoT: to offer instant contact to data at any moment and any location. Figure 7 shows the different types of sensors and their positions in the human body. Table 1 shows the different kinds of body sensor, and Table 2 describes the difference between wearable sensor and implant sensor [18]. Fig. 7 Different types of sensors and their positions in the human body

Remote Patient Monitoring Using IoT, Cloud Computing and AI

61

Table 1 Typical kind of sensor in BSN Type of sensor

Function

Lifetime

ECG

Checking the heart electrical activity

Weeks

EEG

Checking the brain electrical activity

Weeks

EMG

Checking the electrical muscle activity

Weeks

Physiological sensor (Wearable)

Pulse oximeter

Checking O2 concentration in blood

Weeks

Blood pressure sensor

Checking blood pressure

Weeks

Body temperature sensor

Checking body temperature

Weeks

Glucose sensor

Checking glucose concentration in the blood

Years

Drug delivery

Deliver drugs to a target area

Years

Cardiac defibrillator

Correct cardiac arrhythmias

Years

Neuro stimulator

Treatment of movement disorders

Years

Artificial retina

Vision recovery

Years

Endoscope capsule

Visualization of the digestive tract

Hours

Accelerometer

Checking body posture

Weeks

Gyroscope

Checking body orientation

Weeks

Temperature sensors

Checking ambient temperature

Weeks

Humidity sensor

Checking ambient humidity

Weeks

Allergic sensor

Checking allergic agent in the air

Weeks

Implantable sensors

Kinetic sensors (Wearable)

Ambient sensors (Wearable)

Table 2 Difference between wearable and implantable sensors

Characteristics

Wearable sensor

Implantable sensor

Volume

Larger

Smaller

Weight

Heavier

Lighter

Power

High

Low

Operation time

Several weeks

Several years

Recharge ability

High

Low

Biocompatibility

Low

High

Example

Patch, watch, and T-shirt type

Capsule-type

62

M. V. V. P. Kantipudi et al.

3 Cloud Infrastructure and Processing 3.1 Overview Cloud computing provides online information for foundation, stockpiling, and application [19] to control and design. This part gives the distinctive geography of cloud computing for correspondence. The fundamental idea about cloud handling includes parts, models of distributed computing, and their essential qualities. Furthermore, it plots cloud framework and their administration models. At last, this part gives ongoing innovation utilized in cloud computing.

3.2 Topology and Network Protocol for Remote Monitoring Peer-to-peer Inter-Cloud federation: Clouds work together legitimately with one another yet may utilize circulated substances for registries or expediting, as shown in Fig. 8. Mists speak with one another and haggle legitimately without intermediaries [20]. Centralized Inter-Cloud federation: Clouds utilize a focal substance to perform or encourage asset sharing. The focal substance goes about as a storage facility where the accessible cloud assets are enrolled. Concentrated inter-cloud federation is portrayed in Fig. 9. Multi-cloud Service: Customers access the various cloud through assistance facilitated by the cloud customer either remotely or in-house, as shown in Fig. 10. The administrations contain agent segments. Multi-cloud Libraries: Clients build up their own intermediaries by utilizing a brought together cloud API as a library, as shown in Fig. 11. The utilization libraries encourage the use of mists uniformly between clouds. Fig. 8 Peer-to-peer inter-cloud federation [20]

Remote Patient Monitoring Using IoT, Cloud Computing and AI

63

Fig. 9 Centralized inter-cloud federation [20]

Fig. 10 Multi-cloud service [20]

Fig. 11 Multi-cloud libraries [20]

With the fast expansion in distributed computing applications and asset prerequisites, cloud worker remaining task at hand and unpredictability are expanding rapidly, the designs of cloud worker framework become increasingly perplexing. Checking and the board for the distributed computing framework are turning out to be increasingly troublesome. Intelligent platform management interface (IPMI) convention can be applied to customary workers the executives. As it may, colossal CPU, organization, and capacity assets carry special tests to remote administration of cloud worker resource [21]. To addresses the issues of far-off checking the board, in 1997,

64

M. V. V. P. Kantipudi et al.

Intel, HP, Dell, NEC four organizations began to create the executive’s norms, they made a primer IPMI particular in 1998, IPMI2.0 adaptation was accessible in 2004, it incorporates extra encryption, verification, virtual local area network (VLAN), serial over LAN (SOL), and different capacities. It was in reverse viable with 1.0 and 1.5 variants of the detail; it gave more prominent security, freedom, adaptability, and offer-of-band the executive’s ability.

3.3 Cloud Infrastructure Cloud computing design is arranged into two standard areas: front end and back end, as shown in Fig. 12. The customers act as front-end users, utilizing cloud supervision and organization of workers with any PC program as a back end-user. Figure 13 defines the three service models of cloud computing. Software as a Service (SAAS)—In the first model of services, a request is facilitated as support of a client who gets to it through the web and a few models of this category are mail administrations, information sharing, Google docs, and web examination [20]. Platform as a Service (PAAS)—It supplies all the assets needed to fabricate applications, and there is no compelling reason to download programming; otherwise, it is called a cloud product. Administrations gave by PAAS are application plan, improvement, testing, sending, and facilitating. Google App Engine and Microsoft Azure are two models in this classification.

Fig. 12 Cloud computing architecture

Remote Patient Monitoring Using IoT, Cloud Computing and AI

65

Fig. 13 Cloud computing defines three service models, widely known as ‘as a service’

Infrastructure as a Service (IAAS)—It gives the equipment on which application gave by SAAS and PAAS can take a shot at. So, it is otherwise called hardware as a service (HAAS).

3.4 Cloud Computing Components and Characteristics Components of Cloud—Customers, information centers, and circulated servers are the three primary cloud computing mechanisms. Clients—Clients have smartphones through which they are communicating their information on the cloud. Data Center—It is a hub of computers where users can subscribe to the consumers through the Internet. Distributed Servers—It will provide an alternate server if one server fails to provide the services. It will increase the scalability while providing the services. Essential Characteristics—The cloud computing is describing five essential characteristics as per NIST. Rapid Elasticity—It will provide the up and down scalability resources for cloud computing. Measured Service—Cloud providers will monitor the cloud services, i.e., billing, optimization of resources, planning, etc. On-Demand Self-Service—Cloud computing will provide the services without any human being.

66

M. V. V. P. Kantipudi et al.

Ubiquitous Network Access—It will provide excellent services throughout the network for thick and thin clients. Resource Pooling—Multi-tenant model will provide the cloud services as per the demand of clients via physical or virtual resources. Cloud models—Cloud computing will provide excellent services regarding security issues or privacy.

4 Challenges in Cloud and AI-Based IoT on Remote Monitoring 4.1 Overview Cloud computing performs a crucial part in the progressive world and facilitates the various functions from structure to societal media. Such scheme must survive with changing load and developing use signifying humanities’ collaboration and enslavement on computerized processing methods while gratifying excellence of provision (EoP) assurances. Facilitating these techniques is a legion of intangible methodologies integrated to meet the challenge of advancing computing functions. There is a necessity to detect new methodology to realize present and upcoming challenges in cloud computing. This section aims to deliver how emerging technologies such as IoT, AI, and blockchain influences the cloud computing system.

4.2 Accessing Cloud with Validation The cloud validation service is a confirmation stage with an obscure crossover design, and it will control the client’s access assets with unified admittance and confirmation approaches. The cloud validation service incorporates straightforward and intuitive verification techniques for multifaceted personality affirmation. These techniques incorporate biometric strategies, such as unique mark confirmation, equipment gadgets, RSA SecurID Token and FIDO authenticators, and setting-based validation utilizing elements, such as the client’s area and organization. Trust in a client’s personality can likewise be set up through danger examination, because of client attributes, for example, past conduct, authenticators recently utilized for validation, and different elements. Figure 14 shows the functions of the cloud authentication service. A cloud validation service arrangement comprises four primary segments: the Cloud Authentication Service, the Identity Router, the Cloud Administration Console, and the RSA SecurID Authenticate application introduced on client gadgets. The cloud validation service will perform real-time resources for the protected resources. It will improve the validation capability and services run on Microsoft Azure, a cloud computing platform, etc. For all clients, services are standard because

Remote Patient Monitoring Using IoT, Cloud Computing and AI

67

Fig. 14 Cloud authentication service

of their shared nature. An identity router is a virtual appliance that communicates with the components such as cloud authentication service, identity sources, and RSA authentication manager. The cloud validation service components contain a facilitated, multi-inhabitant cloud administration console that SuperAdmins use to perform arrangements and the executive’s undertakings [22].

4.3 Block Chain-Oriented Healthcare Records Blockchain methodology is the latest innovation in intelligence computing machinery. It operates an absolute disseminated ledger of deals accumulated in a chain up of cell blocks that may be manipulated to track record deals. Blockchain developed an extensively recognized idea with the starter of bitcoin. In common, a blockchain setup has three distinct kinds of nodes such as straightforward node, full node, and mining node. A specific node is efficient in delivering and accepting deals in the system and does not store up a ledger copy. A full node delivering justifies all the deals in the system and stores a copy of the ledger. A full node has the capability of mining. Mining is well-known as the procedure of creating a new block in the system [23]. In real time, smart e-health functions can watch patient data by accumulating data from implantable devices and wearable sensors creating private area networks. Smart

68

M. V. V. P. Kantipudi et al.

gateways accumulate and operate local managing of information accumulated from appliances, incorporating noise filtering from health devices, information compression and combination and examines permitting discovery of critical movements in the patient’s health. For example, a quick gateway accumulating test samples from a pacemaker may enhance its sample proceeding to a heart attack, discovered via premanaging at the fog level. As generally edge machines have calculated and storage capacity limitations, also difficult restraints occur in delivering an optimum method for information security and preserving reliability. To guarantee the information is safeguarded, blockchain methodology has been implemented in the IoT field and more real-time techniques. The crucial idea of blockchain meets from numerous disputes while incorporating with fog processing frameworks. Storage space capacity and scalability are considered due to elevated cost and preservation overheads. Though only complete nodes store the complete chain, the storage capacity constraints are considerable. Also, the blockchain liabilities still involve fog frameworks. Efficient agreement processes want to be established to authenticate blocks and restricted the distribution and replication of blocks [24].

4.4 Reliability and Complexity in Computational Intelligence CC is a model that empowers pervasive, helpful, and on-request network admittance to a mutual pool of configurable registering assets. It is conceivable to get to both programming and equipment models distantly in CC applications and practically no information about their physical or consistent areas. Because of its low sending and the executive’s costs, the CC worldview is in effect progressively utilized in a wide assortment of online administrations and applications, including distant calculation, programming as-a-administration, off-web page stockpiling, amusement, and correspondence stages. In any case, a few parts of CC applications, for example, framework plan, advancement, and security issues have gotten too mind-boggling to even think about being effectively treated utilizing conventional algorithmic methodologies under the undeniably high unpredictability and execution requests of current applications. CC assets may not be satisfactory because of their portable applications, generally far from clients topographically. Edge computing (EC) disseminates the applications with low-inertness and high-unwavering quality prerequisites. As both CC and EC are asset touchy, a few significant issues emerge, such as leading work planning, asset designation, and assignment offloading, impacting the entire framework’s presentation. Numerous improvement issues have been defined to handle these issues. These improvement issues ordinarily have complex properties, which may not be tended to by the customary curved advancement-based arrangements [25]. Computational insight (CI), comprising of a bunch of nature-roused computational methodologies, as of late shows incredible potential intending to these advancement issues in CC and EC. As of late, computational intelligence (CI) methods have

Remote Patient Monitoring Using IoT, Cloud Computing and AI

69

cultivated competent answers for CC applications. CI strategies, for example, fake neural organizations, profound learning, fluffy rationale, and transformative calculations, have empowered improving CC ideal models through their capacities of extricating information from high amounts of correct information, in this way further streamlining their plan, execution, and security concerning conventional methods [26].

5 Case Studies 5.1 IoT-Based Remote Pain Monitoring System: From Device to Cloud Platform [27] Facial appearances are among social indications of agony that can build up a programmed human agony appraisal apparatus. Such an instrument can be an option in contrast to oneself report strategy and primarily serves patients. A smart gadget is proposed to cover the patient’s surface electromyogram (sEMG) in this effort. The wearable gadget fills in as a remote sensor hub and is coordinated into an IoT framework for far-off torment observing. Furthermore, both low energy utilization and wearing solace are thought of through the wearable gadget plan for long-haul checking. A versatile web application is created for real-time spilling of high-volume sEMG information, computerized signal handling, deciphering, and perception to show ongoing agony information to parental figures distantly. The IoT-based distant agony observing application and its design have appeared in Fig. 15, which is intended for the real-time focal checking of inpatients and concentrated consideration of patients in the medical clinic. It is made from four sections: wearable sensor node, gateway, cloud sensor, and mobile web application.

Fig. 15 Distant pain observing in the clinic for inpatients and ICU patients

70

M. V. V. P. Kantipudi et al.

Fig. 16 Structure of IoT-based biopotential estimation framework and cloud-based information stream in the programmed pain observing framework

Gateway is the intermediate between sensor hubs and cloud, and the gateway can be an overall switch, the individual hotspot of a cell phone, or a shrewd passage supporting with added highlights, for example, heterogeneity, adaptability, and unwavering quality. The framework can profit from intelligent gateways, mainly when heterogeneous information and correspondence innovations exist in the general medical care far-off the observing framework. In this design, an HTML5-based portable application goes about as an intelligent interface among framework and parental figures. It can deliver continuous waveform, lead lightweight calculations, spare information to an in-program information base and synchronize with the far-off information base workers, as appeared in Fig. 16.

5.2 Internet of Things Sensor Assisted Security and Quality Analysis for Healthcare Datasets Using Artificial Intelligent-Based Heuristic Health Management System [28] In this work, an artificial health management framework has been planned and created. This technology provides more security in the patient dataset and therapeutic administrations’ relationship over its various perspectives. These administrations incorporate the limit for masters, specialists, orderlies, and staff to make better choices quicker. Using IoT helped fake shrewd-based heuristic health management frameworks improve and limit the security hazard on medical care informational collections with IoT sensors. An IoT sensor gathered the clinical information from different assets. The dataset gathered from 10 subjects while performing 12 specific actual exercises (e.g., standing inert, sitting, and unwinding, resting, strolling, climbing/plummeting steps, midsection twists forward, the frontal rise of arms, knees bowing, cycling, running, running, and hop front and back) as shown in Fig. 17.

Remote Patient Monitoring Using IoT, Cloud Computing and AI

71

Fig. 17 M-health data assortment measure structure

Among the three sensors, chest-related sensors giving the ECG estimations are used to decide the various arrhythmias and assess the overall heart observing cycle. The explanation behind picking the IoT cell phone well-being information has its capacity to access the well-being data from any patient anyplace, simple to infection finding, essential patient data, and viably convey the preparation on portable well-being information. As per the portrayal, the IoT sensor was put as portrayed in Fig. 17.

5.3 A Survey on Deep Transfer Learning to Edge Computing for Mitigating the COVID-19 Pandemic [29] Worldwide health at times faces pandemics as is as of now confronting COVID-19 infection. The dispersion and contamination components of this illness are high. Countless individuals from most nations are tainted within a half year from its first report of appearance and continue spreading. The necessary frameworks are not prepared up to certain stages for any pandemic; subsequently, moderation with the existing limit gets essential. On the other hand, deep learning (DL) could moderate COVID-19 like pandemics regarding stop spread, analysis of the sickness, drug and immunization revelation, therapy, quiet consideration, and some more. However, this DL requires enormous datasets just as unique processing assets. These sorts of gear make the frameworks modern and robotized, which assists with adapting to a flareup. Be that as it may, these are outfitted with low registering assets; along these lines,

72

M. V. V. P. Kantipudi et al.

Fig. 18 Structure for edge computing

applying DL is likewise somewhat testing; consequently, DTL additionally would be successful there. In Fig. 18, a standard current COVID-19 episodes circumstance and conceivable working model have appeared. In the primary situation, we may utilize a few medical care sensors like circulatory strain sensors, internal heat-level sensors, webcam, and so on to detect the information about every patient’s running medical issue. At that point, all of the gathered information would be distributed off the EC layer. If the produced report is an essential ailment, a programmed framework-ready message will be sent with subtleties to the medical clinic control room and all related group specialists. In the second situation, a few public spot-checking sensors (like a robot, CCTV, traffic cameras, and so forth) could be utilized to recognize pointless unlawful swarm with or without wearing PPE with assistance from the DTL-EC-based model. On the off chance, the model finds assembling, at that point, a programmed framework-ready message will be sent with all the subtleties to the close-by power.

Remote Patient Monitoring Using IoT, Cloud Computing and AI

73

References 1. Sepanos, D., & Wlof, M. (2018) Internet-of-Things (IoT) system: Architectures, algorithms, methodologies. Springer International Publishing. 2. Marques, G., Bhoi, A. K., de Albuquerque, V. H. C., K.S., H. (Eds.) (2021). IoT in healthcare and ambient assisted living. Springer. 3. Ahsan, M. M., Gupta, K. D., Nag, A. K., Poudyal, S., Kouzani, A. Z., & Mahmud, M. A. P. (2020). Applications and evaluations of bio-inspired approaches in cloud security: A review. IEEE Access, 8, 180799–180814. 4. Saran, P., Rajesh, D., Pamnani, H., Kumar, Hemant, S. T. G., & Shridevi, S. (2020). A survey on health care facilities by cloud computing. In Proceedings of the 2020 International Conference on Emerging Trends in Information Technology and Engineering (ic-ETITE) (pp. 1–5), Vellore, India. 5. Rajkumar, K., & Dhanakoti, V. (2020). Methodological survey to improve the secure data storage in cloud computing. In Proceedings of the 2020 International Conference on Emerging Smart Computing and Informatics (ESCI) (pp. 313–317), Pune, India. 6. Jagirdar, Reddy, & Qyser. (2014). Cloud computing basics. International Journal of Advanced Research in Computer and Communication Engineering, 1(5), 1–6. 7. Yeasmin, S. (2019). Benefits of artificial intelligence in medicine. In Proceedings of the 2nd International Conference on Computer Applications & Information Security, ICCAIS’2019 (pp. 1–6). 8. Oliveira, Lopes, & Govcopp. (2020). What can we expect from the future? The impact of artificial intelligence on society. In Proceedings of the 15th Iberian Conference on Information Systems and Technologies (CISTI) (pp. 1–6). 9. Mohammed, Z. (2019). Artificial intelligence definition, ethics and standards. In Electronics and communications: Law, standards and practice. The British University in Egypt Publications. 10. Fast, E., & Horvitz, E. (2017). Long-term trends in the public perception of artificial intelligence. In Proceedings of the 31st AAAI Conference on Artificial Intelligence (pp. 963–969). 11. Zhong, G., Zhang, K., Wei, H., Zheng, Y., & Dong, J. (2019). Marginal deep architecture: Stacking feature learning modules to build deep learning models. IEEE Access, 7, 30220– 30233. 12. Negra, R., Jemili, I., & Belghith, A. (2016). Wireless body area networks: Applications and technologies. Procedia Computer Science, 83, 1274–1281. 13. Roudjane, M., & Messaddeq, Y. (2020). Innovative wearable sensors based on hybrid materials for real-time breath monitoring. Journal of Wireless Sensor Networks—Design, Deployment and Applications, 1, 1–22. 14. Islam, S. U., Ahmed, G., Shahid, M., Hassan, N., Riaz, M., Jan, H., & Shakeel, A. (2017). Implanted wireless body area networks: Energy management, specific absorption rate and safety aspects. In Chapter 2, Ambient assisted living and enhanced living environments (pp. 17–26). Elsevier. 15. Roy, G., Bhoi, A. K., & Bhaumik, S. (2021). A comparative approach for MI-based EEG signals classification using energy, power and entropy. IRBM. 16. Pramanik, M., Pradhan, R., Nandy, P., Bhoi, A. K., & Barsocchi, P. (2021). Machine learning methods with decision forests for Parkinson’s detection. Applied Sciences, 11(2), 581. 17. Bhatt, T. V., Patel, R. K., Chitara, H. B., Marques, G., & Bhoi, A. K. (2020). Fuzzy logic system for diabetic eye morbidity prediction. International Journal of Computer Applications in Technology, 64(4), 339–348. 18. Bhoi, A. K., Sherpa, K. S., & Khandelwal, B. (2018). Arrhythmia and ischemia classification and clustering using QRS-ST-T (QT) analysis of electrocardiogram. Cluster Computing, 21(1), 1033–1044. 19. Wang, Y., Wang, H., Xuan, J., Dennis, Y. C., & Leung, D. Y. C. (2020). Powering future body sensor network systems: A review of power sources. Journal of biosensors and Bioelectronics, 166, 1–23.

74

M. V. V. P. Kantipudi et al.

20. COV 2020: Cloud computing overview. https://www.tutorialspoint.com/cloud_computing/ cloud_computing_overview.html [5 Nov 2020]. 21. Rani, K., Rani, P., & Babu, V. (2015). Cloud computing and inter-clouds—Types, topologies and research issues. In Proceedings of the 2nd International Symposium on Big Data and Cloud Computing (ISBCC’15) (Vol. 50, pp. 24–29). 22. Hongsong, C., & Xiaomei, M. (2015). Design and implementation of cloud server remote management system based on IMPI protocol. In Proceedings of the 2015 IEEE 12th International Conference on Ubiquitous Intelligence and Computing and 2015 IEEE 12th International Conference on Autonomic and Trusted Computing and 2015 IEEE 15th International Conference on Scalable Computing and Communications and Its Associated Workshops (UIC-ATC-ScalCom) (pp. 1475–1478), Beijing. 23. Dinesha., H. A., & Agrawal., V. K. (2012). Multi-level authentication technique for accessing cloud services. In Proceedings of the 2012 International Conference on Computing, Communication and Applications (pp. 1–4), Dindigul, Tamilnadu. 24. Ismail, L., & Materwala, H. (2020). Block—A blockchain-based framework for health records management. Proceedings of ICCMS, 164–168. 25. Gill, S. S., Tuli, M., Xu, M., Singh, I., Singh, K. V., Lindsay, D., Tuli, S., Smirnova, D., Singh, M., Jain, U., Pervaiz, H., Sehgal, H. B., Kaila, S. S., Misra, S., Aslanpour, M. S., Mehta, H., Stankovski, V., & Garraghan, P. (2020). Transformative effects of IoT, blockchain and artificial intelligence on cloud computing: Evolution, vision, trends and open challenges. Journal of Internet of Things, 10, 1–30. 26. Asim, M., &Wang, Y. (2020). A review on computational intelligence techniques in cloud and edge computing. IEEE Transactions on Emerging Topics in Computational Intelligence, 1–22. 27. Labati, Genovese, Piuri, Scotti, & Vishwakarma. (2020). Computational intelligence in cloud computing. In Proceedings of Conference on Recent Advances in Intelligent Engineering, Topics in Intelligent Engineering and Informatics (pp. 111–126). 28. Yang, G., Jiang, M., Ouyang, W., Ji, G., Xie, G., Rahmani, A. M., Lijeberg, P., & Tenhunen, H. (2018). IoT-based remote pain monitoring system: From device to cloud platform. IEEE Journal of Biomedical and Health Informatics, 22(6). 29. Amoon., M., Altameem., T.,& Altameem., A. (2020). The Internet of Things sensor assisted security and quality analysis for health care data sets using an artificial intelligence-based heuristic health management system. Journal of Measurement, 161, 1–9. 30. Sufian, A., Ghosh, A., Sadiq, A. S., Smarandache, F. (2020). A survey on deep transfer learning to edge computing for mitigating the COVID-19 pandemic. Journal of Systems Architecture, 108, 1–11. 31. Meera, A. J., Kantipudi, M. P., & Aluvalu, R. (2019). Intrusion detection system for the IoT: A comprehensive review. In International Conference on Soft Computing and Pattern Recognition (pp. 235–243). Springer, Cham. 32. Subhadra Bose Shaw, A. K. Singh. (2014). A survey on cloud computing, 2014 International conference on green computing communication and electrical engineering (ICGCCEE). 33. Gill, S. S., Shreshth Tuli, M. Xu, Inderpreet S., et al. (2019). Transformative effects of IoT, Blockchain and Artificial Intelligence on cloud computing: Evolution, vision, trends and open challenges, Internet of Things. 34. Juan A. Añel, Diego P. Montes, Javier Rodeiro Iglesias. (2020). Chapter 3 From the Beginning to the Future, Springer Science and Business Media LLC. 35. Karthik, S., Kumar, S., Prasad, K. M., Mysurareddy, K., & Seshu, B. D. (2020). Automated Home-Based Physiotherapy. In 2020 International Conference on Decision Aid Sciences and Application (DASA) (pp. 854–859). IEEE. 36. Prasad, K. M., Pradeep, K. N., Kashyap, S. S., & Anusha, V. S. (2021). Time Series Data Analysis using MachineLearning-(ML) Approach. Library Philosophy and Practice, 1–7.

An Analytical Study of the Role of M-IoT in Healthcare Domain Bidisha Chanda, Pradeep Kumar Mallick, and Gyoo-Soo Chae

Abstract This study describes mobile health in context of Internet of Things. The Internet of Things applied in development of healthcare systems have reached an evolutionary process. It describes how IoT was introduced in healthcare. It mainly focuses on how mobile IoT gained prominence in the healthcare and medical sectors. This chapter brings to light several applications of M-IoT in healthcare in measuring body temperature, monitoring blood glucose level, ECG, etc. The effect of Internet of Things has been evolving in every aspect of living; however, the effect it has at the healthcare industry is massive because of its growing need and accuracy. The Internet of Things will become extra dominant, and it will be accompanied by the capabilities of mobile computing. It enhances the capability of IoT in healthcare surroundings via means of bringing a large aid within the scope of mobile fitness (m-fitness). Mobile computing assists IoT packages in healthcare, contributes to the present day within the healthcare industry. Mobile computing provides aids to IoT applications in healthcare. It has also mentioned about various benefits and advantages of this system. Security is very critical in IoT-based m-healthcare systems. So this study also includes a section that addresses the privacy related issues, security, and challenges of the M-IoT system. Overall integration of M-IoT with healthcare will drastically reduce healthcare costs and unnecessary hospitalizations. Keywords Internet of Things (IoT) · Mobile IoT · Healthcare · Blood pressure · Regression

B. Chanda · P. K. Mallick (B) School of Computer Engineering, Kalinga Institute of Industrial Technology, Deemed to be University, Bhubaneswar 751024, India B. Chanda e-mail: [email protected] P. K. Mallick · G.-S. Chae Division of Smart IT Engineering, Baekseok University, Cheonan-si, South Korea e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 A. Kumar Bhoi et al. (eds.), Hybrid Artificial Intelligence and IoT in Healthcare, Intelligent Systems Reference Library 209, https://doi.org/10.1007/978-981-16-2972-3_4

75

76

B. Chanda et al.

1 Introduction to IoT and Healthcare Conventional healthcare and medical services are based on the technique of operating with common and qualitative statistics, which no longer work well. The drastic boom in population has introduced many demanding situations within the healthcare device which in turn has made the prevailing healthcare services inefficient. This is where Internet Of Things comes to the rescue [1]. A network of gadgets and different devices, embedded with sensors, electronics, and community connectivity and software which permits those gadgets to gather and transfer statistics is what we call Internet of Things. The effect of IoT within the discipline of scientific research is significant. IoT has enabled advent of larger and scientific treatments. This guarantees that the patients may be monitored from their homes, which reduces the haste to visit doctors. Various gadgets use IOT to make development within the quality of the healthcare and clinical services acquired by the patients [2]. IoT-based healthcare offerings target at imparting rich consumer satisfaction at low cost. From clinical perspective, actual time statistics might permit quicker and rational decision making. The Internet of Things (IoT) has made it feasible to carry a world of opportunities within the discipline of medication and healthcare services. Since the involvement of IOT in healthcare services humans has extra control on their lives and remedies because it includes actual time statistics. It has been anticipated through Cisco Systems that by 2020, IOT will encompass almost 50 billion gadgets linked to the Internet. These gadgets consist of smart devices like actuators, mobile phones, and computer systems that have the potential to switch statistics in a secured way [3]. A form of gadgets along with RFID tags, clinical gadgets, cell phones, and so on accommodates the surroundings of Internet of Things. These devices are related through particular identifiers and interact with each other. The transfer of facts among numerous IOT devices results in an ultra-modern derived fact. Healthcare monitoring devices and wireless communications and IoT infrastructure provide considerable improvement within the field of healthcare and medical offerings [4]. There are classifications of uses of IOT and below listed are ten such fields where the use of Internet of Things is widely used. • Participatory sensing: It involves mobile system which encourages users to maintain records and distribute information and thus helping in recreation of advanced knowledge. • Eco-feedback: It makes the use of mobile applications that generates feedback for the customers about various environmental events or about their personal consumption. • Actuation and control: Applications like mobile application and other mobile applications that control some physical movements or actuates some instances. • Health: Mobile applications are used to get health status of the sensors that are attached to the patient body. These applications update the doctors regarding the current status of the patients.

An Analytical Study of the Role of M-IoT in Healthcare Domain

77

• Sports: Includes an aggregate of bodily sensors and cellular programs which can be used all through game activities to file diverse metrics and assist to enhance the overall performance of the athlete user. • Farming: Related to smart farming practices to enhance productivity, control of farm animals and increase consumer delight and transparency. • Gaming: This class is about digital video games which remember the physical presence or fame of the mobile user to decorate the gaming experience. • Transportation: A developing area in which the sensing capabilities of the cellular smartphone are harnessed for better experience and comfort of parking. • This is the wider category because it pertains to efforts which prioritizes interacting with web-enabled physical entities to be had with inside the nearby environment, as an instance tagging technologies. • Social interaction with people: Mobile applications play an important role on interacting with the everyday lives of the people by interacting with them socially. The first section of this chapter is about introduction of IoT. The next section elaborates integration of M-IOT in healthcare. It describes how the M-IOT has been integrated with healthcare so that it can offer improved healthcare facilities and patients will enjoy these facilities at low cost. The subsequent segment describes about the operating precept of M-IOT and the way the IOT tracks gadgets and assist to preserve the actual time facts. Certain current technology within the area of m-IoT and m-fitness has additionally been highlighted within the following segment which offers an assessment among the exceptional technology. The next section offers with the diverse packages of IOT in healthcare. The next part is all about literature surveys. The seventh segment offers a case look at on diverse advantages of IoT in healthcare. The subsequent segment deals with disadvantages, demanding situations and protection issues of IoT and M-IoT in healthcare. It offers the demanding situations of IOT in information just like the security of information. Next segment has case studies. At the end of this study, the future scope of IOT in healthcare is in short discussed [4] (Table 1).

2 Integration of M-IOT in Healthcare The new technological tendencies related to cellular computing and its applications make it affordable and sufficient to combine it with healthcare and clinical services. The main idea behind the use of mobile IOT in healthcare sectors is to obtain better and error-free readings by making effective and efficient utilization of all possible limited and available resources so as to provide quality services to the patients.. The cellular computing impacts the IoT with the aid of using imparting diverse offerings, programs, and gadgets. These smart devices are successful to screen numerous fitness frameworks like blood pressure, degree of sugar in blood, and pulse rate. These mobile fitness gadgets, for the reason of doing those tasks, are related to the IoT server. The use of mobile devices in collecting and tracking real-time data from patients and

78

B. Chanda et al.

Table 1 Sensing technologies at each category of IOT/WOT-based mobile applications Category

Popular technologies used

Types of data analysis

Market value

Participatory sensing

GPS, camera, microphone

Location-based search, map visualization, information sharing

N/A

Actuation and control

Smart electricity meters, smart appliance, light switches, smart factory

Rule based inference, adaptive reasoning, optimization

$1.3–4.6 Billions

Health

Body area sensors, GPS, microphone, accelerometer, communication, and conservation sensing

Personalization profiling, anomaly detection, emotion detection, ML

$110–900 Billions

Sports

Body area sensors, motion sensors, GPS, pressure shoes, pedometers

Personalization profiling, anomaly detection, historical comparisons, performance visualization, statistics

$200–450 Billions

Agriculture

Wearable collars, GPS, barcodes, RFID tags

Historical comparisons, $140–200 Billions stream data processing, retrieval of info

Gaming

Accelerometer, camera, pedometer, barcodes

Activity recognition, machine learning, information sharing

Transportation

GPS, compass, career Location-based search, connectivity, RFID big data analysis, tags stream data processing

$500–740 Billions

Interaction with things

NFC technologies, RFID tags, barcodes, QR codes

Location-based search, information retrieval, information sharing

$70–150 Billions

Social interaction with people

Bluetooth, PS, camera, microphone

Location-based search, information retrieval, information sharing

$170–450 Billions

Eco-feedback

Camera, energy monitors, smart meters, barcodes, environmental sensors, e.g., air quality

Historical comparisons, $200–750 Billions stream data processing, retrieval of info

$450–635 Billions

An Analytical Study of the Role of M-IoT in Healthcare Domain

79

storing it on the Internet is referred to as m-health [5]. This stored data can be collected by mixed group of clients like the doctors, medical staff, health insurance companies, hospitals, etc. The prime goal of m-health is to obtain information immediately to diagnose illness, keep a track of diseases, take urgent medications for emergency cases. Mobile health is especially important in rural areas where it becomes difficult for the doctors as well as the patients to maintain regular clinical visits. Below given are some healthcare utilization trends: (1) (2) (3) (4)

It is believed by more than two-thirds of healthcare technology executives that use of m-health will have great impact in the future According to health IT outcomes, global m-health revenue will reach $49.12 billion by 2020. It is believed by 40% of surveyed physicians that m-health applications such as tele-health could cut down on a number of clinical visits. According to 93% of the doctors m-health apps can improve patient’s health and having a m-health application connected to emergency health services is valuable.

Mobile fitness will develop through creating country-based e-Health techniques that include it into the present e-Health domain. Policies want to be complemented with the aid of using standards, architectures, and stable partnerships to assist mhealth tasks mature and comprehend their complete potential—utilizing cell and Wi-Fi technology to enhance fitness and well-being. Two important instances are mfitness and cellular learning (m-learning). Most m-fitness programs are utilized by the healthcare specialists for diverse duties like analysis of disease, drug reference, and clinical calculations. Utilizing IoT-MD to screen continual ailments is a brand new approach of gathering a step forward in clinical records. Internal video display units and wearable sensors provide a regular collection of records to people and clinical providers, which alert for adjustments in fitness situations even as constructing a clinical profile for destiny healthcare needs. Cell telephones and cellular gadgets also are adapting to IoT-MD through medically authorized attachments and apps used to screen blood samples or provide reminders to carry out improvement exercises. Figure 1 gives a graphical representation of the global market size share of M-IOT in heal in dollars. It presents a clear picture of what percentage of mobile IOT is used in various healthcare domains.

3 Working Principle of M-IOT in Healthcare The precept of m-IoT is primarily based on tracking gadgets. These devices track the patient’s condition remotely after which send this precise records to a back-end device or a cloud based device which stores it. Then the back-end device minutely examines and analyzes the records and generates the precise indications. The clinicians then come across the disease according to the acquired alert and may effortlessly determine any case of emergency and approximately the sort of remedy this is for use

80

B. Chanda et al.

Fig. 1 Application categories

for the treatment [6, 27–35]. The tracking tool needs not be a scientific tool; however, it may additionally be gadgets or devices like smart watch or a few sort of cell telephones that can hook to the human beings. Recently cell telephones have emerged to be the handiest IOT gadgets. IoT gadgets accumulate critical records and switch that records to medical doctors for real-time tracking and notifies human beings approximately the severity of the disorder through cellular apps and lots of different gadgets connected with it. Medical gadgets sense, reveal, and transmit private records via the gadgets to steady clouds or platforms in which findings can also additionally be analyzed to create actionable, records-related measures and recommendations. Figure 2 gives a diagrammatic representation about the working principle of M-IoT in healthcare. Here, all the data collected from various sources like hospitals, laboratories, etc, are stored in a back-end system or the clouds. All the databases are stored in the cloud. While examining a patient, a doctor needs to retrieve the data about the patient from the database itself. The doctor writes a prescription using the Eprescribing system which is again stored in the clouds. On the other hand, whenever required by the doctors, a pharmacist accesses the database to get details about the prescribed medicines. The hospital can use the back-end system or the cloud to store patient health records as well as details regarding contacting the patents for care.

An Analytical Study of the Role of M-IoT in Healthcare Domain Fig. 2 Global IoT in healthcare market size, by application

81

Sales 11%

20%

12% 19%

21% 17%

Telemedicine

Medication Management

Clinical operations

Patient monitoring

Connected imaging

Others

4 Comparative Analysis of Existing M-IoT Technologies Adaptive Network Technology(ANT) It is an ultra-low-power, short-variety Wi-Fi generation that has the potential to permit gadgets to run on a battery for years. It is a multicast Wi-Fi sensor generation that lets in the tracking gadgets to connect with everyone effortlessly in sensible and precious ways. The ANT protocol is the generation that powers the health ANT+sport, health and health product ecosystem. It has the potential to aid AES-128 encrypted channels for stable connections and may join topologies like peer-to-peer, star, tree, and mesh typologies. These gadgets can also additionally use freq between 2400 and 2524 MHz; however, 2457 MHz is excludeds. Its number one purpose is to permit clinical and sports activities sensors which allows conversation with a show unit, for instance, an eye or cycle computer. Zigbee This generation is solely designed to cope with the specific desires of low-value, low-power Wi-Fi IoT networks, the sensors and manage gadgets that employ lowvalue connectivity. These Zigbee’s WPANs perform at 868 MHz, 902–928 MHz, and 2.4 Ghz frequencies inside 10-a hundred meter inside range. Zigbee is able to controlling and monitoring. Zigbee networks may be prolonged with the usage of routers and permit many nodes to interconnect with every other for constructing an extensive region network. Zigbee can join topologies like point-to-point, point-to-multipoint, and mesh networks. Zigbee operates on IEE 802.15. four radio specification. Zigbee has many programs in healthcare monitoring. Near field communication(NFC) Near field communication (NFC) is a contactless communication technology primarily based on a radio frequency (RF) that uses a base frequency of 13.56 MHz. NFC era is flawlessly designed to change statistics among gadgets through an easy contact gesture that makes use of magnetic discipline induction to permit conversation among gadgets when they are touched together or delivered inside some centimeters of every other. The NFC tag may be incorporated

82

B. Chanda et al.

with smart mobile phones and can be used as a digital fitness file to keep the patient’s clinical history, prescriptions and real-time sensor data. Wi-Fi Wi-Fi is known to be the most power efficient. It is best known for ease of deploymentt. Wi-Fi connections can be integrated with mobile phone to provide much better and efficient healthcare services. Standard Wi-Fi, being the plain preference for IoT, has barriers in each variety and electricity efficiency. These shortcomings was addressed by IEEE for publishing specifications for 802.11ah and 802.11ax. Bluetooth Bluetooth makes use of short-wavelength UHF waves among 2402 and 2480 MHz. Bluetooth helps each star topology in addition to mesh topology. It additionally helps factor to factor, one-to-many, many-to-many topologies. If the low-energy requirements, theoretically limitless scalability, and self-recovery reliability of emerging Bluetooth mesh networks are given, well-designed Bluetooth may be incorporated in cell to develop applications that are powerful for such indoor-asset-tracking scenarios. But the identical Bluetooth findings could be gold standard for accomplishing different commercial enterprise goals. Smart tracking gadgets may be related with Bluetooth can assist the clinicians to asset tracking. Table 2 presents a comparison between different technologies of mobile IoT. The comparison is based on various factors like frequency, channels, modulation, complexity, power profile, extendibility, etc. There are other technologies like LPWANs, RFID. Only small blocks of data can be sent by LPWANs at much lower rate so they are suited for use cases that do not require high bandwidth and are not time-sensitive. Compared to LPWAN, Zigbee ensures higher rates of data transfer. As it uses mesh configuration, it is much less power efficient. Table 2 Comparison of different M-IoT technologies Low-energy bluetooth

Zigbee

NFC

Low power Wi-Fi

Frequency (MHZ)

2402–2482

2402–2482 902–928

13.56

2400–2500

Channels

3

1

3

Modulation

GFSK

BPSK and QPSK

ASK

64 QAM

Max potential data rate

1.1 Mbps

251 Kbps

25 Kbps

55 Mbps

16

Range

11 m

101+m

11 cm

31 m

Power profile

Days

Month/year

Month/year

Hours

Complexity

Complex

Simple

Simple

Complex

Nodes/master

8

65001

1+1

–

Extendibility

No

Yes

No

Yes

An Analytical Study of the Role of M-IoT in Healthcare Domain

83

Radio frequency identification (RFID) works using radio waves for transmitting small amounts of data from an RFID tag to a reader within a very short distance.

5 Applications of M-IOT Devices in Healthcare Continuous glucose monitoring(CGM) An advanced way that monitors glucose readings is continuous glucose monitoring. The glucose level in the patients’ blood is monitored by the CGM device. The connection of the tool is made to an insulin pump with an automated suspension of insulin infusion. It connects to an analytic platform to monitor health data over time to recognize patterns/triggers for abnormal levels. There are systems that collect and store glucose readings in hospital database system, which in turn makes it easy for clinical staff and the doctors to access real-time data from their patients. Statistics have demonstrated that CGM may reduce patient’s long-term complications between 40 and 75% of the time. These gadgets are computerized and may file blood strain for 24/7, while sufferers can do their regular each day activities. A photoplethysmographic (PPG) primarily based technique is used to screen blood strain. The use of wearable sensor is provided by wearable, cuff-less PPG-based blood pressure monitor with novel height sensor. Thus, IoT applications can help to generate correct glucose level reading with minimized error. Electrocardiogram(ECG) An easy test that diagnoses heart disease is electrocardiogram (ECG). A normal heart’s activity tracking with the aid of using a hand-held ECG tool is powerful and affords long-term financial savings for heart patients. Existing electrocardiogram (ECG)tracking gadgets which include Holter are not convenient for long-time period use because of their length and twist in the wires. Wireless ECG tracking gadgets that could hook up with IoT server are rising into market. A wearable ECG tracking and systems that provide continuous alerts like the “I-Heart” that constantly video display units ECG is advanced. The I-Heart video display units affected person’s ECG and troubles an alert to the affected person if it identifies odd behavior of heart. The tool makes use of a Wi-Fi ECG sensor and a smart phone. Thus integrating IoT technologies heart behavior can be easily tracked with accuracy and minimized errors. Body Temperature Monitoring MosChip has designed an IoT- based hardware and software program for healthcare structures which guide the healthcare employees and caretakers to record their patients’ fitness situations with the help of the Internet remotely. The temperature patch is hooked up to the area of the frame to make sure of the correct temperature. The records are amassed from a sensor and dispatched to the telephone over Bluetooth 4.0. The smart software designed for iOS and Android structures will carry out a record evaluation and ship real-time records to the unique cloud software via over

84

B. Chanda et al.

a GPRS/Wi-Fi backbone. The medical doctors can examine the records and endorse medicine in real-time. The software consists of fitness reviews for faraway diagnosis, alarm onset temperature limit, agenda indicators to take medicine, and scheduled medical doctor visits. Another technology is the use of Arduino in measuring the body temperature. It uses sensors connected to the Arduino board Other than these, there are several other applications of IoT in healthcare like blood pressure monitoring, oxygen saturation monitoring, ECG monitoring, etc. These applications when integrated with the healthcare sectors prove to be efficient and accuracy.

6 Benefits of M-IoT Minimized errors—IOT checks risk of errors. It also deals with precise and exact collection of data, automated workflows and minimized waste. IoT technologies are good at minimizing the human mistakes without removing human touch. Suffering from exhaustion from long hours of work, medical staffs often commit medicalrelated mistakes. This is where IoT comes to the rescue. Better patient experience—A connected healthcare system satisfies the needs of each and every patient. Committed procedures, improved treatment options, and upgraded diagnosis accuracy make for a better patient experience. Patients get satisfied with the quality of medical facilities they receive and also with the efficiency of the technologies as well as the accuracy. Reduced expenses—With IoT, patient no longer needs to be physically present in a clinic for visits, instead monitoring can be done in real-time. Connected home care facilities will also bring down the expenses related to hospital stays and readmissions. Thus, this efficient autonomous system can reduce expenses when it comes to patient cost savings because of fewer health center trips in addition to elevated diagnostics and treatment. Improved ailment management—With real-time facts, healthcare carriers can constantly screen patients. This way that they are able to spot any disorder earlier than it spreads and turns into serious. IoT promises in helping to improve the health of the patients with chronic diseases. It provides an error-free monitoring system using sensors, gateways etc. Remote tracking—Real-time far off tracking through related IoT gadgets and clever signals can diagnose illnesses, deal with sicknesses and shop lives in case of a clinical emergency. With the help of remote monitoring and virtual visits, IoT technologies keep patients better connected to the doctors. Prevention of diseases—Smart sensors examine fitness conditions, way of life selections and the surroundings and advocate preventative measures, a good way to lessen

An Analytical Study of the Role of M-IoT in Healthcare Domain

85

the prevalence of sicknesses and acute states. Frequent monitoring by the IoT devices helps in keeping a track of the present condition of the body. Accessibility—Doctors can view all of the essential records on command and take a look at real-time affected person situations without leaving their office. The database is stored in the cloud from where the doctors can access the required data anytime. Medical facts accessibility—Accessibility of digital scientific statistics permits sufferers to get hold of fine care and assist healthcare vendors make the proper scientific choices and save you complications. Improved remedy management—IoT gadgets assist the management of medicine and the identifies the remedy and decrease clinical error. IoT has plethora of benefits which ensures healthcare system to be more reliable, fast, and quickly accessible. Improved healthcare Management—Using IoT gadgets, healthcare government can get precious statistics approximately gadget and workforce effectiveness and use it to signify innovations. For studies purpose—IOT gadgets have an excessive potentiality for clinical research purposes And they are capable of collect and analyze a big number of useful data.

7 Challenges of M-IoT Still there are many issues that need to be checked before the concept of IoT is widely accepted. It is claimed by Atzoria et al. that people will not take risk by IoT adoptions as long as there is no public confidence that IoT will not cause serious threats and violations to their privacy. Security issues—There are instances where the systems get hacked. There is a dire need to focus on the data security and unauthorized access. There is a risk that fraud people may also get entry to centralized structures and recognize a few merciless intentions. Worldwide healthcare regulations—The government medical establishments that are planning to integrate IOT in their workflow should follow the guidelines that are already issued by international health administrations Data overload and oversight—Studies show that aggregation of data is not eassy due to the use of different communication protocols and standards. Inspite of data overload, IoT devices still have an enormous number of data. The statistics gathered through IoT gadgets are applied to benefit important insights. However, the quantity of statistics is so superb that deriving insights from it have become extraordinarily tough for doctors which, in the long run impacts the decision making

86

B. Chanda et al.

Human error—Exchange of information among medical staff can introduce human errors, which can end up in risking the life of patient. Communicating with patients who lacks verbal communication skills is another great challenge, and information collection problems are even more challenging. Integration of multiple devices and rules—Integration of a couple of gadgets additionally causes hassle in the working of IoT within-side the healthcare sector. The motive for this disturbance is that tool producers have not reached a consensus concerning communique protocols and standard. Existing software infrastructure is outmoded—IT infrastructures in many hospitals are obsolete. They do not support the idea of integrating the IOT devices for betterment of healthcare services. Therefore, IoT in healthcare has to face a lot of challenges and complexities before producing the better and efficient results. No doubts if the complexities can be reduced using certain methodologies, in Fig. 3, a graphical representation of factors contributing to the challenges is presented (Fig. 4).

Fig. 3 Working of M-IOT

An Analytical Study of the Role of M-IoT in Healthcare Domain

87

Fig. 4 Categories of issues contributing to challenges

8 Relevant Studies on M-IoT in Healthcare Skin disease detection from the image processing is a crucial subject technique, however additionally for computational intelligence and image processing. There also exist studies indicating the pores and skin tumors for chemical evaluation. In [7], the chapter mainly focused on how plasmacytoid dendritic cells, Langerhans cells and inflammatory dendritic epidermal cells, influence anti-viral pores and skin protection technique. Here, the main theme of the chapter mentioned the diagnosis which is preferred for pores and skin cancer detection primarily based on the evaluation of auto-antibodies and mucous membranes including usage of commercialized tools and immunofluorescence microscopy. The authors provided how the antigens influence pores and for serological analysis. In [8], the chapter mainly mentioned a technique to locate atopic dermatitis, psoriasis, and call dermatitis characterized with the aid of using transcriptomic profiling. Potential sickness become detected with the aid of using the protein expression levels. In [9], a method was proposed for cancer pores and skin cancers detection. Authors mentioned a device which integrates deep learning knowledge of with pores and skin lesions ensemble technique. In [10], the study provided optoacoustic dermoscopy version primarily depends on a feature of excitation power and perforation intensity measures for pores and skin evaluation with the aid of using extreme imaging. It became proven how to research absorption spectra at more than one wavelengths for visualization of morphological and practical pores and skin features. A method to locate pores and skin cells and cancer primarily based on deep learning knowledge of version become accepted in [11]. The authors have mentioned segmentation with regards to the middle of cancer with the aid of using lesion indexing network which makes use of fully convolutional residual network as a chief detection technique. In [12], we are able to discover an updated evaluation over diverse gadget gaining knowledge of techniques carried out to pores

88

B. Chanda et al.

and skin analysis and detection. While in [13], we are able to discover an extensive contrast of cell apps for pores and skin tracking and cancer detection. The authors have drawn a comparison between numerous applications, mentioned their ability in efficient image processing and sensing technology and additionally provided elements of ethical, pleasant, and obvious improvement of apps for scientific use. Ahn et al. [14] mentioned a device by which phsiological alerts can be measured which includes ECG and BCG. It makes use of a smart chair that is capable of sensing non-limited bio-alerts. Almotiri et al. [15] themed a chapter which discuss about the devices of m-fitness that takes the help of m-gadgets to gather real-time records from sufferers and stores it on network servers. This records may become useful for the scientific checking of sufferers and is attained by taking help of some of wearable gadgets and frame sensor community. Barger et al. [16] designed a smart residence facility that enables the usage of a sensor community to reveal and analyze the movements of the patient and a prototype of the equal is likewise being tested. The first and the most important goal in their work is to test if their device is succesful to outsmart the behavioral styles and feature mentioned. Chiuchisan et al. [17] deviced a framework to eradicate the spread of diseases from the affected person in smart ICUs. Dwivedi et al. [18] evolved a framework in an effort to stable the scientific records that needs to be communicated over the Internet for Electronic Patient Record (EPR) structures wherein they suggest a multi-layered architecture of healthcare records device framework that is an aggregate of public key infrastructure, smartcard, and biometrics technology. Gupta et al. [19] proposed a version which measure the fitness of the affected person the usage of Raspberry Pi and may be of an extraordinary usage for the medical sectors and sufferers in addition to their own circle of relative’s members. Gupta et al. [20] introduce a method that makes use of Intel Galeleo improvement board that gathers the different records and updates it to the database from wherein it is accessed by the doctors

9 Discussion of the Use M-IoT in Cancer Detection Cancer is the out of control increase of skin cells. It develops when an unpaired DNA damages the pores and skin cells due to exposure to the ultraviolet radiation of the tanning beds and sun, causing mutations (genetic defects) which leads to the skin cells to multiply swiftly ensuing the formation of malignant tumors. Sometimes, the pores and skin cancers may even unfold and harm the nearest cells [21]. Also, in a few cases, pores and skin most cancers might also additionally develop on vital organs. [22] Sun exposure may be referred to as one of the fundamental purpose for improvement of tumors in pores and skin cells, even though there are numerous different elements like surroundings threats, radiation evaluation, or even inheritance can play a role. The Internet of Things and data and generation carried out in improvement of healthcare and clinical structures have reached an evolutionary process. Nowadays, skin related cancers result in the demise of a huge population which is consistent with the healthcare assessment system. To take a look at the cost of

An Analytical Study of the Role of M-IoT in Healthcare Domain

89

the treatment, there is a demand for a computerized device to assess an affected person’s risk of cancers. The discussed system includes the detection of pores and skin cancer with excessive degree of accuracy and rapid execution time with the utilization of support vector system and particle swarm optimization algorithm and moreover developing the Twitter Doctors Community if you want to transfer the affected person’s health related information to the individuals in their family the use of mobileular phone and different clever devices or doctors with the assist of Arduino. Additionally, doctor can suggest appropriate remedy for the patient and can interact with sufferers with the aid of IoT and moreover enhancing and making sure safety-based healthcare system in which affected person’s statistics bases are measured in a strong manner with the use of Anti SQL injection to save server statistics from being hacked. Discussed Methodology: The discussed methods describe how skin cancer can be detected with more accuracy and efficiency by implementing IoT techniques. These methods include the following methodologies: Image Acquisition: Image acquisition dermoscopic images are basically digital images of magnified pores and skin lesion, captured with traditional camera well equipped with unique lens extension. The lens connected to the dermatoscope act like a microscope magnifier with its very mild source that illuminates the pores and skin layers evenly. Dermatologist can create accurate documentation of amassed photographs, maintaining a sequence of steps for analysis, wherein photographs are processed that allows you to extract records that can later used to categorize those images. Pre-Processing Techniques: Pre-processing of image is an important step that permits you to reduce noises and improve the standard of the image. It is important to be implemented to limit the inside aspect the background having an effect on of the result. The main motive of this step is to enhance the standard of cancer images via a way of method of casting-off unrelated and surplus elements inside aspect the background of the image for in addition image pre-processing. Good choice of pre-processing strategies can substantially enhance the accuracy of the system. In segmentation techniques, the process of segmentation includes to the separation of an image into disjoint regions that are uniform with respect to some property such as color, luminance, and its texture. Thresholding-based segmentation encompasses bi-stage and multi-thresholding. Thresholding technique includes: (1) Histogram and (2) adaptive thresholding. In region-based segmentation, the image is splitted into smaller and smaller additives and then merged into sub-images which can be adjoining and comparable in a few sense. It consists of statistical location merging, multi-scale location. Feature Extraction Using PSO Algorithm: In automatic analysis of pores and skin lesions, characteristic extraction is primarily based on the so-referred as ABCDrule of dermatoscopy. The extracted parameter is optimized to the use of particle swarm optimization algorithm. Particle swarm optimization (PSO) is a stochastic international optimization technique evolved through Eberhart and Kennedy in 1995.

90

B. Chanda et al.

Fig. 5 Block diagram of the discussed methodology

In PSO, a set of solutions traverse the problem space with a speed primarily based on their own experience and the experience of their neighbors. During every spherical of traversal, the speed, thereby the location of the particle, is up-to-date primarily based totally at the above parameters. This technique is repeated until a most beneficial answer is obtained (Fig. 5). Table 3 provides a contrast among the readings taken through the conventional method as opposed to readings measured with the proposed technology. This effortlessly concludes the efficiency, correctness and acccuracy of the proposed method. Observation: This version proposes approximately a stable IoT primarily based totally pores and skin most cancers detection system. Compared to the traditional bioscopic approach, this proposed approach proves to be more efficient. The diagnosing method uses superior strategies like the digital image processing techniques combined with support vector machine similarly to the particle swarm optimization algorithm for the class of the photograph from normal skin photograph. The consequences confirmed accuracy. The alert concerning the illnesses may be despatched trough twitter alert or mail alert. Through ESP8266 Wi-Fi module, the database is despatched to the IoT platform. By the use of SVM_PSO algorithm, the cancerous cell is separated from healthy pores and skin.

An Analytical Study of the Role of M-IoT in Healthcare Domain Table 3 Comparison of readings between traditional approach of skin cancer detection versus discussed methodology

91

1

18

12

2

20

15

3

25

21

4

28

21

5

32

26

6

17

13

7

22

17

8

30

25

9

28

22

10

25

20

11

23

18

12

19

15

13

20

16

14

23

18

15

26

22

16

33

28

17

16

13

18

27

22

19

32

28

10 M-IoT in Blood Pressure Measurement from Heart Rate High blood pressure also called hypertension will increase the risk of mortality for the adults. [23–26] Generally, high blood pressure is described as blood strain above 140/ninety, and it will become extreme if the strain is above 180/120. A latest observation has proven that people of China who have an excessive heart rate in among 80–90 bpm are much more likely to have a short lifestyles in comparison to the humans with heart rate among 60–69 bpm. Thus, a linear relationship exists among heat rate and blood pressure. Systolic blood pressure measured from the heart rate based on IOT is discussed that makes use of LM358 sensor and NRF module to calculate the blood strain the usage of regression techniques. Devices and components: Some of the components and sensors used in constructing the devices are Arduino Uno(AT328), heart rate sensor(LM328), NRF module, LCD display, and power supply. This device aims at calculating blood pressure from measured heart rate. Arduino Uno(AT328) works as the main processing unit for the proposed device. The arrangements for the components have been shown in the given figure. The power supply block sends the power supply to the device. The sensor used for the heart rate measurement records the heart rate data and transfers it

92

B. Chanda et al.

to the Arduino Uno for further processing followed by which the transmitted information is sent serially to the NRF module, and the readings of the blood pressure and heart rate are shown on the LCD. Discussed Methodology: In this discussed methodology, more than some data sets were collected from students within age group of 20–28. Figure 6 is a block diagram that represents the arrangement of devices in the discussed methodology. Their corresponding blood pressure and heart rate values were collected from the heart rate sensor via Aurdino Uno. Figure illustrates the discussed regression model. At first, the information of heart rate and its corresponding blood pressure is gathered after which divided the data set into three classes inclusive of excessive blood pressure, excessive blood pressure, ordinary blood pressure. The degrees for every class are given in Table 4. Based on the pattern shown, a regression model is generated in Table 4. This table is developed to show regression model for each class. From the generated regression pattern, equations are made which are enforced on the device to get the measure of blood pressure from the heart rate. In Fig. 7, the block diagram represents the arrangement to obtain the regression pattern (Tables 5 and 6). It may be visible from Table 7 that the distinction among blood pressure and blood strain calculated with the aid of using the proposed version may be very minimal. This offers accuracy to be 99% accurate. On the alternative hand, comparable statement has been made for class 2 and class three sufferers. The accuracy rate of class 2 sufferers has been 90% while that of class three sufferers turned into discovered to be 89%. Table gives the effectiveness of the proposed version. Thus, this approach proves to be correct in phrases of the calculations and accurate readings of the blood strain. Fig. 6 Heart rate measure by using this device

Table 4 Category of blood pressure ranges

Category/class

Blood pressure (BP)

Range

First category/class 1

High BP

BP>=140

Second category/class 2

Pre high BP

120