Tele-Healthcare E-Book

• Herausgeber: John Wiley & Sons
• Kategorie: Wissenschaft und neue Technologien
• Sprache: Englisch

Seitenzahl: 519

Veröffentlichungsjahr: 2022

Beschreibung

TELE-HEALTHCARE

This book elucidates all aspects of tele-healthcare which is the application of AI, soft computing, digital information, and communication technologies, to provide services remotely and manage one’s healthcare.

Throughout the world, there are huge developing crises with respect to healthcare workforce shortages, as well as a growing burden of chronic diseases. As a result, e-health has become one of the fastest-growing service areas in the medical sector. E-health supports and ensures the availability of proper healthcare, public health, and health education services at a distance and in remote places. For the sector to grow and meet the need of the marketplace, e-health applications have become one of the fastest growing areas of research. However, to grow at a larger scale requires the following:

1. The availability of user cases for the exact identification of problems that need to be visualized.
2. A well-supported market that can promote and adopt the e-health care concept.
3. Development of cost-effectiveness applications and technologies for successful implementation of e-health at a larger scale.

This book mainly focuses on these three points for the development and implementation of e-health services globally.

In this book the reader will find:

• Details of the challenges in promoting and implementing the telehealth industry.
• How to expand a globalized agenda of personalized telehealth in integrative medical treatment for disease diagnosis and its industrial transformation.
• How to design machine learning techniques for improving the tele-healthcare system.

Audience

Researchers and post-graduate students in biomedical engineering, artificial intelligence, and information technology; medical doctors and practitioners and industry experts in the healthcare sector; healthcare sector network administrators.

Leseprobe

Table of Contents

Cover

Title Page

Copyright

Preface

1 Machine Learning-Assisted Remote Patient Monitoring with Data Analytics

1.1 Introduction

1.2 Literature Survey

1.3 Machine Learning in RPM

1.4 System Architecture

1.5 Results

1.6 Future Enhancement

1.7 Conclusion

References

2 A Survey on Recent Computer-Aided Diagnosis for Detecting Diabetic Retinopathy

2.1 Introduction

2.2 Diabetic Retinopathy

2.3 Overview of DL Models

2.4 Data Set

2.5 Performance Metrics

2.6 Literature Survey

2.7 Discussion and Future Directions

2.8 Conclusion

References

3 A New Improved Cryptography Method-Based e-Health Application in Cloud Computing Environment

3.1 Introduction

3.2 Motivation

3.3 Related Works

3.4 Challenges

3.5 Proposed Work

3.6 Proposed Algorithm for Encryption

3.7 Algorithm for Decryption

3.8 Experiment and Result

3.9 Conclusion

References

4 Cutaneous Disease Optimization Using Teledermatology Underresourced Clinics

4.1 Introduction

4.2 Materials and Methods

4.3 Proposed System

4.4 Challenges

4.5 Results and Discussion

References

5 Cognitive Assessment Based on Eye Tracking Using Device-Embedded Cameras via Tele-Neuropsychology

5.1 Introduction

5.2 Materials and Methods

5.3 Framework Elements

5.4 Proposed System

5.5 Subjects

5.6 Methodology

5.7 Results

5.8 Discussion

5.9 Conclusion

References

6 Fuzzy-Based Patient Health Monitoring System

6.1 Introduction

6.2 System Design

6.3 Software Architecture

6.4 Results and Discussion

6.5 Conclusions and Future Work

References

7 Artificial Intelligence: A Key for Detecting COVID-19 Using Chest Radiography

7.1 Introduction

7.2 Related Work

7.3 Materials and Methods

7.4 Experiment and Result

7.5 Results

7.6 Conclusion

References

8 An Efficient IoT Framework for Patient Monitoring and Predicting Heart Disease Based on Machine Learning Algorithms

8.1 Introduction

8.2 Literature Survey

8.3 Machine Learning Algorithms

8.4 Problem Statement

8.5 Proposed Work

8.6 Performance Analysis and Evaluation

8.7 Conclusion

References

9 BABW: Biometric-Based Authentication Using DWT and FFNN

9.1 Introduction

9.2 Literature Survey

9.3 BABW: Biometric Authentication Using Brain Waves

9.4 Results and Discussion

9.5 Conclusion

References

10 Autism Screening Tools With Machine Learning and Deep Learning Methods: A Review

10.1 Introduction

10.2 Autism Screening Methods

10.3 Machine Learning in ASD Screening and Diagnosis

10.4 DL in ASD Diagnosis

10.5 Conclusion

References

11 Drug Target Module Mining Using Biological Multifunctional Score-Based Coclustering

11.1 Introduction

11.2 Literature Study

11.3 Materials and Methods

11.4 Proposed Approach: MR-CoC

multi

11.5 Experimental Analysis

11.6 Discussion

11.7 Conclusion

Acknowledgment

References

12 The Ascendant Role of Machine Learning Algorithms in the Prediction of Breast Cancer and Treatment Using Telehealth

12.1 Introduction

12.2 Literature Review

12.3 Architecture Design and Implementation

12.4 Results and Discussion

12.5 Conclusion

12.6 Future Work

References

13 Remote Patient Monitoring: Data Sharing and Prediction Using Machine Learning

13.1 Introduction

13.2 Literature Survey

13.3 Problem Statement

13.4 Machine Learning

13.5 Proposed System

13.6 Results and Discussions

13.7 Privacy and Security Challenges

13.8 Conclusions and Future Enhancement

References

14 Investigations on Machine Learning Models to Envisage Coronavirus in Patients

14.1 Introduction

14.2 Categories of ML Algorithms in Healthcare

14.3 Why ML to Fight COVID-19? Tools and Techniques

14.4 Highlights of ML Algorithms Under Consideration

14.5 Experimentation and Investigation

14.6 Comparative Analysis of the Algorithms

14.7 Scope of Enhancement for Better Investigation

References

15 Healthcare Informatics: Emerging Trends, Challenges, and Analysis of Medical Imaging

15.1 Emerging Trends and Challenges in Healthcare Informatics

15.2 Performance Analysis of Medical Image Compression Using Wavelet Functions

15.3 Results and Discussion

15.4 Conclusion

References

Index

Also of Interest

End User License Agreement

Guide

Cover

Table of Contents

Title Page

Copyright

Preface

Begin Reading

Index

Also of Interest

End User License Agreement

Pages

v

ii

iii

iv

xv

xvi

xvii

xviii

xix

xx

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

159

160

161

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

383

384

385

386

387

388

389

391

392

393

394

Scrivener Publishing

100 Cummings Center, Suite 541J

Beverly, MA 01915-6106

Artificial Intelligence and Soft Computing for Industrial Transformation

Series Editor: Dr. S. Balamurugan ([email protected])

Scope: Artificial Intelligence and Soft Computing Techniques play an impeccable role in industrial transformation. The topics to be covered in this book series include Artificial Intelligence, Machine Learning, Deep Learning, Neural Networks, Fuzzy Logic, Genetic Algorithms, Particle Swarm Optimization, Evolutionary Algorithms, Nature Inspired Algorithms, Simulated Annealing, Metaheuristics, Cuckoo Search, Firefly Optimization, Bio-inspired Algorithms, Ant Colony Optimization, Heuristic Search Techniques, Reinforcement Learning, Inductive Learning, Statistical Learning, Supervised and Unsupervised Learning, Association Learning and Clustering, Reasoning, Support Vector Machine, Differential Evolution Algorithms, Expert Systems, Neuro Fuzzy Hybrid Systems, Genetic Neuro Hybrid Systems, Genetic Fuzzy Hybrid Systems and other Hybridized Soft Computing Techniques and their applications for Industrial Transformation. The book series is aimed to provide comprehensive handbooks and reference books for the benefit of scientists, research scholars, students and industry professional working towards next generation industrial transformation.

Publishers at Scrivener

Martin Scrivener ([email protected])

Phillip Carmical ([email protected])

Tele-Healthcare

Applications of Artificial Intelligence and Soft Computing Techniques

Edited by

R. Nidhya

Manish Kumar

and

S. Balamurugan

This edition first published 2022 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA © 2022 Scrivener Publishing LLC

For more information about Scrivener publications please visit www.scrivenerpublishing.com.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

Wiley Global Headquarters

111 River Street, Hoboken, NJ 07030, USA

For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.

Limit of Liability/Disclaimer of Warranty

While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials, or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read.

Library of Congress Cataloging-in-Publication Data

ISBN 978-1-119-84176-0

Cover image: Pixabay.Com

Cover design by Russell Richardson

Set in size of 11pt and Minion Pro by Manila Typesetting Company, Makati, Philippines

Printed in the USA

10 9 8 7 6 5 4 3 2 1

Preface

In the current world scenario, public healthcare has become one of the prime importance. Providing a better health service to one and all is the major concern. Scientists around the world working on different Artificial Intelligence and soft computing techniques to provide a better healthcare service. Along with the scientists, Industries also coming forward to transfer the application of soft computing techniques for providing e-health and telehealth service to public. Countries which are densely populated require industrial services like telehealth for day-to-day health issues.

In today's scenario healthcare industry is growing at a greater pace. Telehealth is the application of AI, soft computing, digital information and communication technologies, to provide healthcare services remotely and manage your healthcare. The services may comprise of the technologies which one use in home or those which are suggested or recommended by the doctors to improve or support healthcare services. In general, telehealth or e-health is the industrial transformation of soft computing techniques for the common people. The most basic element of telehealth is telecommunications, which uses a wider range of information and communication technologies (ICTs).

As said earlier, telehealth service is now a day's provided by many industries. Telehealth can also be seen as examples for virtual home healthcare, where patients suffering from chronicle diseases are taken care by the family members with certain procedures and services available at home. Telehealth revolution has made the things easier for those situated at a remote place and is in need of diagnosis, care and referral of patients. Training of proper healthcare or health related issues may sometimes be provided via telehealth schemes or with related technologies such as e-health, which make use of small computers and internet.

Now-a-days soft computing techniques are directly influencing human life in terms of healthcare analysis and management. Soft Computing techniques are currently being used for monitoring and recording patient diseases and their symptoms for proper diagnosis of the same. There are numbers of soft computing techniques which can be used for early detection and prevention of diseases. It may help us in identifying the root cause of the disease and provide a better solution for telehealthcare application. In the era of technology and industrial revolution, many agencies are working to develop a soft computing-based paradigm which can easily identify and monitor the health status of persons situated a distance place within a specified period of time.

There are number of soft computing application that can easily be used in industries to solve telehealthcare problem. These applications include gathering information of patient, medical records and data, intelligent diagnosis and carefulness strategies, detecting disease etc. A Telehealth industry provides an innovative combination of new applications that underlines the advancement in medical fields. Today, the world is in huge humanitarian crises with respect to healthcare workforce shortages, and a growing burden of chronic disease. As a result, telehealth has become one of the fastest-growing service areas in medical sector. Telehealth supports and ensure the availability of proper healthcare services, public health, and health education services at a distance and remote places. It is assumed that telehealth applications has become one of the fastest growing area of research, but to grow at a larger scale it requires the following: 1) the availability of use cases for exact identification of problems that need to be visualized. 2) A well supported market that can promote and adopt the telehealthcare concept. 3) Development of cost-effectiveness applications and technologies for successful implementation of telehealth at a larger scale.

Coverage area of the book, importance and intended audience:

1. The major objective of this book is providing a platform for presenting soft computing-based telehealthcare service. It will cater to the needs of undergraduate and postgraduate students as well as research students. It will also aid academicians, researchers, and industry experts working on healthcare systems backed by soft computing techniques.

2. To enlist the challenges in promoting and implementing telehealth industry.

3. To globalize the agenda of personalized telehealth in integrative medical treatment for disease diagnosis and its industrial transformation.

4. Design of machine learning technique for better implantation of telehealthcare system in current scenario.

In chapter 1, Vinutha et al. has introduced a ML based remote monitoring of patients with Data Analytics. Monitoring the health condition post-discharge or post-operation is required to ensure a speedy recovery. We all know that, healthcare services can be benefited from technological advancements to ensure better service. In the proposed work the authors have collected the patient's data using sensors and uploaded them to the cloud. The collected data is subjected to pre-processing followed by analysis. In the proposed approach, the patient's health is remotely monitored and machine learning techniques are applied to foretell abnormalities in patients' health condition. The suggested method can also be effective during the pandemics like COVID-19 with the scare availability of medical personnel and treatment resources, this prediction may help in taking appropriate measures at the earliest.

In chapter 2, Priyadarshani and Jagdeesh Kannan did an extensive survey for an intelligent system which can used for detecting Diabetic Retinopathy. As we all know that Diabetic retinopathy (DR) is a complication that causes changes in blood vessels present in the retina which leads to vision loss at later stages and has no or lesser symptoms during the initial stages. In this research, authors have summarized and analyzed all the recent computer-aided diagnosis (CAD) systems based on the nature of data, recent approaches taken in classifying DR, performance comparison using statistical parameters, existing limitations, and challenges.

In current scenario the world is in need of e-health services. In chapter 3, Dipesh et al. propsed an e-Health applications based on cloud computing environment that allows healthcare services to monitor the health and health related information of patients dynamically. The main aim of this work is to provide a secure infrastructure for e-Health application hosted in cloud computing environment.

In chapter 4, authors have assessed if tele-dermatology could improve primary care provider (PCP)-delivered care for cutaneous disease at a clinic serving uninsured patients. They respectively analysed all consultations and collected patient age, tele-dermatology diagnosis, time to tele-dermatology reply, time to next dermatology appointment, as well as PCP and tele-dermatologist proposed care plans. By their investigation, authors analysed that expanding tele-dermatology for PCPs in underresourced clinics has the potential to improve treatment of cutaneous disease by non-specialists and to mitigate suboptimal care for disadvantaged patients.

In chapter 5, Shanmugaraja T. et al., a significant issue has been addressed for the ageing population. As there is a lack of medical centres to meet the needs for in-person assessment. Authors suggested remotely distributed evaluations using webcam eye-tracking data, such as visual paired comparison (VPC) tasks, could enhance access to external, intermittent neuropsychological testing for cognitive decline, and can be seen as solution for future e-health application.

In chapter 6, Venkatesh T. et al. presented an approach in which smart devices, Computational Intelligence (CI) and Soft Computing Techniques (SCT) may help the doctors to monitor the data of their patients suffering with various healthcare issues and also to diagnose and to provide state-of-the-art treatment. Applying SCT, identification of correlated features, feature ranking or importance and feature selection are performed on UCI Machine learning datasets and also classification and prediction are performed on the datasets to examine the accuracy of the predictions for the classification algorithms.

In chapter 7, COVID-19 has become a global pandemic in such a short period of time that it should be taken seriously and we all need to be precautious about it at all times. In such times, it is difficult to visit hospitals physically and spend a lot of time there for manual tests. The use of CXR images for diagnosis of COVID-19 proves to be of great help. As the traditional methods used for diagnosis are not enough for clinical purposes, there is a need to make use of artificial intelligence to improve the performance of these methods. During the development of the model, authors realized how complicated the architecture was becoming, and hence tried their best to optimize the whole network to diminish erroneous situations and bring out the best result possible. As the traditional methods used for diagnosis are not enough for clinical purposes, they made use of artificial intelligence to improve the performance of these methods.

In chapter 8, authors have proposed an enhanced IoT framework for patient health monitoring and predicting the heart disease based on Machine Learning Algorithms. Proposed method combines both IoT and machine learning concepts and is used to provide efficient and effective remote health monitoring for patient, determine whether a patient is having a particular disease or not as well as to provide a quick solution to the patients in case of some emergency situation which requires immediate doctor attention. As per the experimental result carried out by the authors it is concluded that naive Bayes is good for handling clinical data.

In chapter 9, authors have addressed the information security and authentication issue using neural network and discrete wavelet transform methods. To provide security and authentication to the data, authors proposed biometric based authentication method. In this method, the brain waves activity is used as a biometric authentication for user recognition. The acquisition of EEG signal is done and compressed using DWT (Discrete Wavelet Transform). The Feed Forward Neural Network (FFNN) is used for pattern matching to provide accurate results than existing methodologies.

Autism Spectrum Disorder (ASD) is the most challenging developmental disorder among the children. In chapter 10, authors were discussed about the various details related to ASD such as how ASD is diagnosed using assessment tools called ASD Screening Tools used by the medical staff with the support of parents/caregiver, improving the accuracy and classification of ASD using computational intelligence, etc. This chapter provides a detailed review of different screening tools and machine learning methods for ASD diagnosis.

The novel drug discovery for most of the complex diseases is based on the drug target proteins. In chapter 11, authors proposed a multi-functional score based co-clustering approach MR-CoCmulti which is introduced for drug target module mining with five novel biological scores namely hydrophobic residues density, sequence length, polar residues density, amino acid density, molecular weight scores. Authors suggested the drug target modules based on the biological functionality and drug target proteins in the results. As per their experimental result the proposed method outperforms than the existing methods.

Telehealth has become a necessary technology in this pandemic situation. So, in chapter 12, authors have proposed the telehealth method for predicting and treating breast cancer using machine learning algorithms such as Logistic regression, K-nearest neighbor, Support Vector machines, Kernel SVM, Decision Tree algorithm and Random forest classification. Based on the experimental results the authors concluded that K-SVM method is predicting the malignant cells with more accuracy than other algorithms. However, after the prediction the patient can be given remote treatment based on the severity of the cancer.

In continuation with previous chapter, Chapter 13 also deals about data sharing and prediction using Machine Learning algorithms for remote patient monitoring. The proposed system recommends an instantaneous solution and provides proper assistance during an emergency, by predicting its seriousness using modern algorithms. The data stored in the database is tested by the KNN classifier and linear regression for better performance. Hence the system proposed by the authors has scalability that deals with the online prediction and facilitates patient care remotely.

In chapter 14, authors have discussed about the different Machine learning models to forecast presence of the coronavirus disease in a patient. This proposed work analyses the foreseen of the diseased people from people with minor indications built on 111 impute relating to medical and the clinical examination facts and performed the prediction and severity analyses using various models such as Naive Bayes, Support Vector Machine, Artificial Neural Network, K-Nearest Neighbour (KNN), Convolutional neural network (CNN), Logistic regression and Decision tree. From the experimentation results it was identified that Artificial Neural Network is the best when compared to other models in terms of Accuracy, Specificity and Precision.

In chapter 15, the authors concentrated about the health informatics. Medical recording is not only an observation or storage it acts as communication medium between physicians and other medical people involve in observing and recoding the patient data. Recorded high standard medical data is a preventive measure of serious disorders from the people in highrisk future. This chapter aims to provide information about current trends, challenges and issues in healthcare informatics and analyzing the performance of various compression techniques in medical image.

The Editors

June 2022

1Machine Learning-Assisted Remote Patient Monitoring with Data Analytics

Vinutha D. C.1*, Kavyashree2 and G. T. Raju3

1Dept. of CS &E (AI&ML), Vidyavardhaka College of Engineering, Mysuru, India

2Dept. of ISE, Vidyavardhaka College of Engineering, Mysuru, India

3Dept. of CSE, Tne Oxford College of Engineering, Bengaluru, India

Abstract

The health condition of the patients needs to be monitored with immense care. Healthcare promotes good health, helps in monitoring the patient's health status, disease diagnosis, and its management along with recovery. Monitoring the health condition postdischarge or postoperation is required to ensure a speedy recovery. Healthcare services can benefit from technological advancements to ensure better service. Healthcare assisted with machine learning techniques plays a significant role in the effective diagnosis of ailments, monitoring patient's health condition, and extend support in taking suitable measures during abnormality. In the proposed work, we collect the patient's data using sensors and upload them to the cloud. The collected data are subjected to preprocessing followed by analysis. The patient's health is remotely monitored, and machine learning techniques are applied to foretell abnormalities in the patient's health condition. Existing remote monitoring systems are not flexible and, hence, may result in an increased number of false positives. We try to reduce unnecessary alerts via machine learning methods and data analytics. Essential attributes like pulse rate, blood pressure, temperature, gender, and cholesterol levels of the patient are taken into consideration while predicting the results. In the time of pandemics, like COVID-19 with the scarce availability of medical personnel and treatment resources, this prediction may help in taking appropriate measures at the earliest. We train the model with the Kaggle Heart Disease UCI data set and test the model with real-time patient data. We apply our model to k nearest neighbor (KNN) and Naive Bayes algorithm. The KNN has performed well over the Naive Bayes algorithm.

Keywords: Remote patient monitoring, machine learning, cardiovascular disease, k-nearest neighbor (KNN), Naive Bayes

1.1 Introduction

Health is the one of the most important factors that has a significant impact in life. In the current era, both internal and external factors contribute to ill health. Heredity aspects, mental status, and lifestyle also have an influence on the health condition. The heart, being a vital organ in the human circulatory system, operates diligently by performing various functions, such as pumping blood, supplying oxygen, and nutrients to all other organs parts of the body. Nowadays, heart-related diseases are observed in individuals of all age groups. Patients undergo diagnosis, treatment, and surgery if necessary. Monitoring the health status of a patient regularly especially postsurgery is crucial. Diagnosing a heart disease is a challenging task that requires expert support and technological assistance. In this regard, remote patient monitoring (RPM) appears as a boon serving distant patients by monitoring their health status continuously.

1.1.1 Traditional Patient Monitoring System

The conventional medical system involves patient consulting a doctor related to a health problem. The doctors suggest necessary medical scans or tests required to diagnose the ailment of the patient whenever essential. Depending on the test results or scans, doctor diagnoses the disease and advices the future course of action, like medicine prescription and surgery, if required. After the course of the medicine or postsurgery, the patient has to undergo check-ups at regular intervals to keep track of the patient's health state. The doctor recommends further action to be taken based on the health condition and observations recorded during the regular checkups.

The abovementioned traditional patient monitoring system has the following limitations:

The patient's health condition is monitored only during the regular checkups and is not monitored continuously.

This method requires physical presence of the patient at the hospital, which may cause inconvenience.

This type of monitoring may not be suitable for diseases, where real-time monitoring abnormality identification is crucial to save lives.

Emergency situations cannot be predicted in prior.

Limited hospital resources may not be sufficient to fulfill the needs of the patients.

1.1.2 Remote Monitoring System

Remote patient monitoring systems provide an attractive solution to overcome these limitations. Remote monitoring system (RMS) accompanied with technological advancements monitors the health status of the patients remotely. Unlike the traditional system where telephone calls were used to obtain patient data, telemonitoring systems offer additional benefits. Technological advancements in information and communication systems have enabled the telemonitoring systems to collect and transmit valuable patient health monitoring data, such as blood pressure, blood sugar level, weight, and electrocardiographic signals through wired and wireless networks.

Remote monitoring system provides the following advantages over the traditional patient monitoring system:

Patient's health status is monitored continuously in real time.

Patient can be at home without the need to travel to hospital for check-ups.

RMS can issue an alert at the time of emergency to hospital personnel to make necessary arrangements (e.g., ambulance service, notifying the doctor, setting up ventilators, etc.).

RMS with continuous analysis can predict the emergency situations in advance.

These predictions can help the hospital personnel to undertake suitable measures to increase the hospital resources.

With many attractive solutions, RMS has its own issues [25] to be addressed.

1.1.3 Challenges in RPM

Remote patient monitoring systems involve electronic gadgets, and additionally, the Internet to connect patients and their healthcare service providers. The patient monitoring systems include measurements from several devices, such as glucometer, weighing machine, and blood pressure monitor. These measured values are transmitted to a backend service via medical hub.

Some of the challenges faced by RPM systems are as follows:

Information reliability: Healthcare professionals expect the measurements from the registered devices at standard condition to be trusted. Care should be taken to ensure that the measurement is that of the registered patient under monitor and not someone else. Malfunctioning device may affect the measurement.

Without reliable information, the efforts made to monitor, analyze, and predict the health condition will not be fruitful. An authentication method proposed in the study of Petkovic [1] tries to address the reliability issue by attaching the patient's identity with the device used to obtain measurement. Device authentication and user authentication methods are employed to ensure user and their device appropriately. Here cryptographic keys are derived from information related to user and device authentication.

Information quality: In RPM, measurements are usually taken using user devices, and the quality of the measurement is not always acceptable. Healthcare providers need to rely on the obtained measurement without having knowledge of user device condition. Manual errors while recording measurements can also influence the process.

Two kinds of metadata mentioned in the study of Petković [1] specify the standard of health-related data collected by the patient using a RPM system. The first kind of metadata is regarding the usage of the device by a patient who collected measurements to ensure that the device was used the way it was intended. The second kind of metadata is pertaining to the device.

User privacy: User data collected need to be managed with utmost care with good security practices, policies, and protocols. The information could be taken care of by arbitrators or mediators, which makes hazard for patients possibly having their information taken. The difficulties are no less intense for emergency clinics, who incorporate the mediator frameworks that could be compromised in security putting the patient's well-being and safety in danger.

Real-time access to data: Transferring the gathered information to the RPM system involves multiple hops. Data from the end-user equipment are accumulated and uploaded to a remote site, such as server, for analysis. If an end user is relying on cellular network, then the information has to travel thru the corresponding network service provider's substructure and then out onto the Internet. This involves multiple hops before reaching the destination. Mobile networks are not always available. They can induce delays and may put life of the patient into danger because of network issues.

High power consumption: The RPM monitors and collects the health status continuously. The collected data are transmitted very often. This may cause high power consumption and may drain the battery in a short time for battery-driven devices. An RPM system with low power consumption is proposed in Noman et al. [3] tries to lower the power consumption in smart phones.

1.2 Literature Survey

Proposed system in Jeya Priyadharsan et al. [2] monitors the patient health remotely using sensor networks. The system consists of hardware components, such as heart rate sensor, temperature sensor, blood pressure sensor, and Raspberry Pi board. The heart rate is measured at every instant. The data gathered from the sensors are stored in the cloud for analytics. These data subjected to analytics help in determining abnormal health condition. k Nearest neighbor (KNN) classifier is employed to classify a patient's health status.

The system in Padmashree et al. [4] aims at improving the healthcare systems by analyzing desirable health factors, such as blood pressure, heart rate, and body temperature, to predict heart issues. The systolic and diastolic blood pressures are taken into account. The work efforts to alert the user with an application. The patients who were previously diagnosed with heart diseases were the users. The application enabled the patient to view their health status on their personal device, like mobile phones, and the doctor could view his patient's data and analyze the data using multilayer perceptron (MLP) algorithm.

A system that tracks and monitors the health condition remotely is proposed in Mohammed Baqer et al. [5] collects the health status of a patient via medical sensor. Along with the data collected from the sensors, the patient's current location is obtained using GPS coordinates. Confidentiality and authentication is ensured via encryption through AES algorithm. During emergencies where the patient's health condition is abnormal, a rescue alarm is issued to notify the medical team for assistance.

Remote electrocardiogram (ECG) monitoring system proposed in Pagadala et al. [6] continuously monitors the health condition, collects the required data using AD8232 sensor, and processes the data using ESP 8266 microcontroller.

A noninvasive framework with spatiotemporal filtering and convolution neural network (CNN) is suggested in Ying et al. [7] to remotely monitor heart rate. The MMSE-HR data set is used for the intended research work. Eulerian Video Magnification (EVM) is used with an intent to obtain desirable features that relate to the information on the heart rate. CNN is used to evaluate HR from the feature image.

A pervasive RPM system operating in four modes was proposed in Chao et al. [8] triggers the healthcare services based on physical status rather than feelings. The monitoring system sends physical signals to remote medical applications in real time. Various physiological and environmental indicators are considered in the monitoring process.

An Internet of Things (IoT)-based RPM system in Fayoumi and Bin Salman [9], along with the vital signs measurement, considers the posture, physical activity, and heart rate during monitoring, generates the patient heart risk report with five levels indicating the magnitude of the risk.

A smart healthcare system discussed in Catarinucci et al. [10] suggests the use of RFID in body-centric systems to collect temperature, humidity, and other information from the patient's existing location. These wireless communication-based RFID systems have the potential to collect and process multichannel data about human behavior in agreement with the power exposure and sanitary guidelines.

The IoT tiered architecture (IoTTA) proposed in Nguyen et al. [11] aims at transforming sensor data into real-time clinical feedback. The work proposes a five-tiered architecture comprising sensing layer that involves necessary devices and sensors for recording health parameters, sending layer offers a procedure to connect and share data. Storing and processing layer stores and processes the data, respectively. Mining and machine learning applications are applied for feature extraction, classification, and regression.

Conventional methods do not provide sufficient warning because they involve determining the variation in weight and warning sign monitoring. This has resulted in implantable devices. The implantable devices [12] can be specifically designed sensors or they can be devices, such as permanent pacemakers, cardiac resynchronization therapy devices, and cardioverter defibrillators, whose role is to identify different symptoms.

1.2.1 Machine Learning Approaches in Patient Monitoring

Machine Learning based Health Monitoring System in Gnana and Anu [13] considers five parameters for monitoring system, such as ECG, pulse rate, pressure, temperature, and position detection by using wearable sensors. The data from sensors are subjected to classification using support vector machine (SVM) classifier with the ability to identify emergency conditions.

Work in Wenfeng et al. [14] uses CNN to attain a unified analysis of ECG recordings and radar data. Unlike the traditional system that is based on arrhythmia classification that lacks capability to manage the motion state and low accuracy, work in the study of Chao et al. [8] has improved performance and achieved good accuracy.

The work in Baucas and Spachos [15] proposes a framework to tackle several issues pertaining to remote monitoring system. High power consumption, data security, and privacy concerns are addressed suitably with sound recording, surveillance capture, and speech classification components. With the intension of reducing the power consumption, data are transferred only upon the detection of abnormality instead of transmitting data continuously 24/7.

The intelligent hybrid remote patient-monitoring model proposed in Hassan et al. [16] is context-aware and uses local components along with cloud-based components. The cloud serves the need for storage and processing of big data. The local portion of the system is used at times when the Internet connectivity is lost or at the times of failure in cloud operations. The information discovery procedure is carried out in this structure vertically through changing low-level information into a greater degree of abstraction. High-level feature provider (HLFP) is used to change raw data from low-level to higher level of abstraction, features selection, and classification.

A framework to tag real-time health-related sensor data, HealthSense is developed in Stuntebeck et al. [17]. This framework transmits the sensor data from the patient to a server. Machine learning technique is employed to analyze the sensor data in the server. The system interacts with the user to assist in classification of desired events, like pain, itching, and so on. Weka classifiers were used in training the system. The system is incorporated with a feedback mechanism to increase the accuracy of the machine learning technique.

A noninvasive RPM system proposed in Zhang et al. [18] predicts the risk of heart failure. Variations in levels of N-terminal prohormone in the blood pressures and bodyweight helps in determining whether a patient is at the risk of heart failure or not using SVM. The intended system develops a scoring method that can determine the risk of HF on a long-term basis.

Machine learning-based model for monitoring cardiovascular disease [19] observes trends of vital signs contextualized with data from clinical databases. The data collected from the sensors are analyzed locally and fed to SVM to monitor extracted features and classify the patient as continued risk and no-longer risk.

Patient monitoring system proposed in Siva Priya et al. [20] collects data, such as temperature and heartbeat, which are collected using sensors and uses KNN classifier to predict the patient health status. The work has a buzzer that gives a signal to the duty nurse available if any abnormal condition is found. If the condition gets even worse, it also sends an email through SMTP protocol from the registered mail to the current duty doctor mail.

An electronicmedical record-wide (EMR-wide) feature-based selection approach along with machine learning patient readmission probability prediction proposed in Shameer et al. [21] considers two classes, namely readmitted and non-readmitted. Significant features that matter the most are united into a model with the help of correlation-based feature selection (CFS) method. The model is trained and tested using the fivefold cross-validation method.

A remote patient multimodular system discussed in Medjahed et al. [22] remote monitoring system working with two databases with the help of which distress situations are predicted in elderly people. This telemonitoring system collects data with sensors. Different machine learning algorithms are applied, and their effects are studied.

An e-health traffic flow classifier in Kathuria et al. [23] utilizes machine learning concept to identify the important features and classify the traffic data. This work efforts to train and test the data set using the combination of genetic algorithm along with binary decision tree to recognize desirable features and assign appropriate priority to every packet in the traffic.

1.3 Machine Learning in RPM

Machine learning finds application in several domains. Machine learning algorithms are commonly used in regression, classification, and prediction. Machine learning algorithms are applicable to the medical field as well. In this section, we give an overview of some of the commonly used machine learning algorithms in patient monitoring systems.

1.3.1 Support Vector Machine

Support vector machines are supervised learning models that can analyze data and perform classification and regression analysis. Support vector machine works by finding the optimal hyperplane that maximizes the margin between the data points. An optimal hyperplane that separating the two classes is shown in the Figure 1.1.

The equation of a hyperplane is given below.

where, w is a vector that is normal to the hyperplane and c is an offset.

Multiple possible hyperplanes separating two classes is shown in Figure 1.2.

Figure 1.1 Hyperplane that maximizes the distance between the classes.

Figure 1.2 Multiple hyperplanes in SVM.

When the data are nonlinearly separable, such data are made nonlinearly separable data using the Kernel Trick.

1.3.2 Decision Tree

Decision tree is a supervised machine learning algorithm. The decision tree makes choices primarily based on the current state of the data. Decision tree can help in creation of a model that can learn from trained data and predict correct class when test data are given. The structure of decision tree is shown in Figure 1.3.

The decision tree algorithm works as follows:

i. chooses the finest attribute using attribute selection method to divide the records;

ii. The chosen attribute or feature is made a decision node and further divides the data set into smaller subgroups;

iii. Step i and ii are repeated recursively to construct the tree for each child until one of the conditions is met.

Decision trees predict the class label for a given record by the method of comparison. The process is initiated at the root node. The attribute value of the root is compared with the attribute value of the given record. Based on the result of the comparison, the process proceeds with the corresponding tree branch. This process continues till a leaf node is encountered. Decision trees predict the label by processing the tree to the point where no more classification is possible on reaching the leaf node.

Figure 1.3 Decision tree.

1.3.3 Random Forest

Random forest is a group of decision trees that operate as a unit. Every tree in the random forest gives out a class prediction. The class with majority vote will be the declared as the final prediction. Low association or correlation among the models is the key. Uncorrelated models can create predictions that are more exact than any of the individual predictions. The working of Random Forest algorithm is shown in Figure 1.4.

1.3.4 Logistic Regression

Logistic regression is a supervised classification algorithm used to predict the probability of a target variable. Logistic regression computes the association between the dependent variable and one or more independent variables by estimating probabilities by means of a logistic function or sigmoid function. The hypothesis of logistic regression tends it to limit the cost function between 0 and 1.

Sigmoid function formula is as follows:

Figure 1.4 Random forest.

1.3.5 Genetic Algorithm

A genetic algorithm is an optimization method that imitates the technique of natural selection. The algorithm begins with an essential or principal population collectively with random chromosomes comprising genes with a collection of zeros (0) or ones (1). Further steps of the algorithm biases people toward the most useful solution through iterative strategies like selection operators, crossover, and mutation.

Genetic Algorithm works as follows:

population of individual solution is randomly initialized;

choose the individuals from the populations randomly;

evaluate the fitness and the individuals are compared according to their fitness;

continue the process till the termination condition is met;

modify the individuals using the following operations;

reproduction, copy an individual lacking change;

crossover, exchange sub structure between two individuals;

mutation, exchange a single unit in an individual at a random position.

1.3.6 Simple Linear Regression

Simple linear regression model is used to evaluate the relationship between two continuous variables.

This regression model has the subsequent assumptions:

Homogeneity of variance: the magnitude of the error in the prediction does not vary drastically across the values of the independent variable.

Independence of observations: statistical sampling methods bring out the findings in data set and find no unseen relationships.

Normality: the data follow a normal distribution.

The dependent and dependent variable have linear relationship between them.

The linear regression model take up a linear relationship among the input variables (x) and the output variable (y). The value of variable y can be determined by the linear combination of input variables (x).

Formula for simple linear regression

y is the predicted value, β0 is the y-intercept of the regression line, β1 is the slope.

Regression with a single input variable x is called a simple linear regression. Regression with multiple input variables is referred to as multiple linear regressions.

1.3.7 KNN Algorithm

The k nearest neighbor (KNN) [24] is supervised learning algorithm suitable for classification and regression problems. The KNN algorithm is simple and easy to understand. The KNN does not require assumptions or parameter tuning. The KNN algorithm depends on labeled information to memorize a work and predict an appropriate label when a new unlabeled information is given. We calculate the distance between the points using Euclidean distance.

where a and b are two points; ai and bi are Euclidean vectors; n is space. Figure 1.5 shows the use of KNN for classifying the data points into one of the 3 classes.

The KNN algorithm assumes that similar things exist in close proximity.

Figure 1.5 KNN classification.

KNN Algorithm

1. load the training data, as well as test data;

2. initialize K to your chosen number of neighbors where K can be only integer;

3. for every point in the test data do the following:

compute the distance between test data and training data, sort the distance in ascending order.

4. select the top K entries from the arranged group;

5. simple majority of the category of nearest neighbor return a label as a prediction value.

1.3.8 Naive Bayes Algorithm

Naive Bayes is a statistical classification technique based on Bayes Theorem. It is a probabilistic supervised machine learning algorithm. Bayes theorem relies on the conditional probability to determine the probability of a hypothesis with previous knowledge. Figure 1.6 shows Naive Bayes classification.

Figure 1.6 Naive Bayes algorithm.

Bayes theorem is given as:

The probability of hypothesis X on the event Y is called as posterior probability denoted by P(X/Y).

The probability of the evidence given that the probability of a hypothesis is true is called likelihood probability and is denoted by P(Y/X).

The probability of hypothesis prior to the observance of evidence is prior probability denoted by P(X).

The probability of evidence is marginal probability denoted by P(Y). Naive Bayes classifier calculates the probability of an event as follows:

Step 1:

For the given class labels compute the prior probability.

Step 2:

For each attribute in each class compute the likelihood probability.

Step 3:

Find the posterior probability by substituting these values in the Bayes formula.

Step 4:

Observe the class with higher probability, the input given belongs to the class having higher probability.

Contributions of Proposed Work