Technological Advancement In Internet Of Medical

Things And Blockchain For Personalized

Healthcare Applications And Use Cases A Prasanth
download
https://ebookbell.com/product/technological-advancement-in-
internet-of-medical-things-and-blockchain-for-personalized-
healthcare-applications-and-use-cases-a-prasanth-56337766

Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

Technological Advancement In Algal Biofuels Production Neha Srivastava

https://ebookbell.com/product/technological-advancement-in-algal-
biofuels-production-neha-srivastava-47361000

Technological Advancement In Instrumentation Human Engineering

Selected Papers From Icmer 2021 Mohd Hasnun Arif Hassan

https://ebookbell.com/product/technological-advancement-in-
instrumentation-human-engineering-selected-papers-from-
icmer-2021-mohd-hasnun-arif-hassan-49158398

Technological Advancement In Mechanical And Automotive Engineering

Proceeding Of International Conference In Mechanical Engineering
Research 2021 Muhammad Yusri Ismail

https://ebookbell.com/product/technological-advancement-in-mechanical-
and-automotive-engineering-proceeding-of-international-conference-in-
mechanical-engineering-research-2021-muhammad-yusri-ismail-50501704

Technological Advancement In Ewaste Management Towards Smart

Sustainable And Intelligent Systems 1st Edition Biswajit Debnath

https://ebookbell.com/product/technological-advancement-in-ewaste-
management-towards-smart-sustainable-and-intelligent-systems-1st-
edition-biswajit-debnath-52768018
Technological Advancement In Clean Energy Production 1st Edition
Amritanshu Shukla

https://ebookbell.com/product/technological-advancement-in-clean-
energy-production-1st-edition-amritanshu-shukla-58439170

Functional Foods Technological Challenges And Advancement In Health

Promotion Edited By Sajad Ahmad Wani Mohamed S Elshikh Mona S
Alwahaibi And Haroon Rashid Naik

https://ebookbell.com/product/functional-foods-technological-
challenges-and-advancement-in-health-promotion-edited-by-sajad-ahmad-
wani-mohamed-s-elshikh-mona-s-alwahaibi-and-haroon-rashid-
naik-52873038

Book Of Readings 2nd Annual School Of Science And Science Education

Conference Federal University Of Technology Minna Date 19th22nd
November 2006 Theme Challenges In Science And Technological
Advancement And Economic Reforms 2nd Ed Y Aiyesimi I N Mogbo G A Okoli
https://ebookbell.com/product/book-of-readings-2nd-annual-school-of-
science-and-science-education-conference-federal-university-of-
technology-minna-date-19th22nd-november-2006-theme-challenges-in-
science-and-technological-advancement-and-economic-reforms-2nd-ed-y-
aiyesimi-i-n-mogbo-g-a-okoli-4182212

Technological Advancements In Construction Selected Papers 1st Edition

Angela Mottaeva

https://ebookbell.com/product/technological-advancements-in-
construction-selected-papers-1st-edition-angela-mottaeva-34497840

Technological Advancements In Biomedicine For Healthcare Applications

1st Edition Jinlong Wu

https://ebookbell.com/product/technological-advancements-in-
biomedicine-for-healthcare-applications-1st-edition-jinlong-wu-4633814
 Technological Advancement in
Internet of Medical Things and
Blockchain for Personalized
Healthcare

Technological Advancement in Internet of Medical Things and Blockchain for

Personalized Healthcare presents an overview of the innovative concepts, technolo-
gies, and various biomedical applications of the Internet of Medical Things (IoMT).
Features:

• Provides insights into smart contracts, healthcare monitoring equipment, and

the next generation of Internet of Things sensors to improve adherence to
chronic disease management programs and patient health.
• Discusses the IoMT for personalized healthcare, security, and privacy issues of
the IoMT in the healthcare sector.
• Elaborates on the opportunities and challenges of blockchain technology in the
healthcare system.
• Focuses on the convergence of the IoMT and blockchain for emerging
personalized healthcare systems.
• Presents techniques and methods to secure IoMT devices to protect them from
cyberattacks.

This book is primarily written for graduate students and academic researchers
working in the fields of computer science and engineering, biomedical engineering,
and electrical engineering.
 Technological
Advancement in Internet
of Medical Things
and Blockchain for
Personalized Healthcare
Applications and Use Cases

Edited by
A. Prasanth, Lakshmi D, Rajesh Kumar Dhanaraj,
Balamurugan Balusamy, and Sherimon P.C
First edition published 2024
by CRC Press
2385 Executive Center Drive, Suite 320, Boca Raton, FL 33431
and by CRC Press
4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
CRC Press is an imprint of Taylor & Francis Group, LLC
© 2024 selection and editorial matter, A. Prasanth, Lakshmi D, D Rajesh Kumar, Balamurugan Balusamy,
and Sherimon P.C; individual chapters, the contributors
Reasonable efforts have been made to publish reliable data and information, but the author and publisher
cannot assume responsibility for the validity of all materials or the consequences of their use. The authors
and publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any future
reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter
invented, including photocopying, microfilming, and recording, or in any information storage or retrieval
system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, access www.copyri​ght.com or
contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-​750-​
8400. For works that are not available on CCC please contact [email protected]
Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used
only for identification and explanation without intent to infringe.
ISBN: 9781032521541 (hbk)
ISBN: 9781032521602 (pbk)
ISBN: 9781003405450 (ebk)
DOI: 10.1201/​9781003405450
Typeset in Times
by Newgen Publishing UK
Contents
About the Editors......................................................................................................vii
Contributors...............................................................................................................ix

Chapter 1 Advanced Enabling Technologies of IoMT in Personalized

Healthcare.............................................................................................. 1
K. Arthi and B. Chidhambararajan

Chapter 2 Deep Learning Interpretation of Biomedical Data in IoMT...............16

M. Senthamil Selvi, P.V. Kavitha, P.V. Deepa, and
Ahmed A. Elngar

Chapter 3 Machine Learning for Decision Support Systems in IoMT................ 37

T. Kokilavani, K. Kannan, K.S. Yamuna, B. Sangeetha,
and M. Vetrivel

Chapter 4 Secure Healthcare Systems: Recent and Future Applications............. 62

K. Umapathy, T. Dineshkumar, M. A. Archana, S. Omkumar,
and W. Boonsong

Chapter 5 Transforming Healthcare Management: Combining

Blockchain, P2P Networks, and Digital Platforms............................. 81
Janhvi Rajyaguru and Lakshmi D

Chapter 6 Convergence of IoMT and Blockchain for Emerging

Personalized Healthcare System: Challenges and Use Cases............. 99
Nguyen Tien Dung, Pham Chien Thang, Ta Thi Nguyet Trang,
and A. Prasanth

Chapter 7 Role of Access Control Mechanism for Blockchain-​Enabled

IoMT in Personalized Healthcare...................................................... 110
Shreya Kakkar, Lakshmi D, and Ranju Yadav

v
vi Contents

Chapter 8 Protecting the Privacy of IoT-​Based Health Records Using

Blockchain Technology..................................................................... 127
P. Mathivanan, D. MohanaPriya, P. Manjula,
and Mariya Ouaissa

Chapter 9 Securing IoMT Devices to Protect the Future of Healthcare

from Rising Cyberattacks.................................................................. 145
V. Saranya, T. Devi, and N. Deepa

Chapter 10 Smart Hand-​Hygiene Compliance and Temperature

Monitoring System to Tackle COVID-​19-​like Pathogens
in Healthcare Institutions..................................................................161
Hari Krishnan Ramachandran, Aravind Krishnaswamy Rangarajan,
and Rajesh Kumar Dhanaraj

Chapter 11 LDS–​LVAT: Lie Detection System–​Layered Voice

Technology........................................................................................ 187
Veena K., Meena K., D. Rajalakshmi, Fathima M.,
and Thamarai Selvi K.

Index....................................................................................................................... 205
About the Editors
Dr. A. Prasanth is currently working as an Assistant Professor at Vel Tech Rangarajan
Dr. Sagunthala R&D Institute of Science and Technology, Chennai, Tamil Nadu,
India. He received the B.E. degree in Electronics and Communication Engineering
, the M.E. degree in Computer Science and Engineering (with specialization in
Networks) and the PhD degree in Information and Communication Engineering, all
from Anna University, Chennai, India. Recently, he has received the Young Scientist
Award from the International Scientist Award 2020 for his excellent research per-
formance. He has also received Researcher of the Year Award 2020 from “2nd
International Business and Academic Excellence Awards (IBAE)” at Delhi, held on
26 December 2020. Moreover, he has received the Young Researcher Award from
the “Institute of Scholars Awards” 2020. He has published more than 45 research
articles in reputed international journals, among which 15 articles are indexed
in SCI and 30 articles are indexed in Scopus. He has published and granted two
patents. Further, he has published more than 12 books under reputed publishers. He
has served as Resource person in 25 Faculty development programmes. Moreover,
he has served as Editorial Board Member in various reputed SCI journals. He has
9 years of teaching experience and his research interests include Internet of Things,
Edge Computing, Cloud Computing, and 5G network.

Lakshmi D is presently working as Senior Associate Professor (Grade 2) in the

School of Computer Science and Engineering at VIT Bhopal University, Madhya
Pradesh, India. Till February 2021, she was designated as an Educational Research
Officer at Vishnu Educational Development and Innovation Centre and Associate
Professor at B V Raju Institute of Technology run by Shri Vishnu Educational Society,
Hyderabad, from 2016 to February 2021. She has been working in the educational
sector since 1998. Her key research focus is on exploring the dynamics of learning,
the dynamics of learner, and classroom dynamics, suitable to accelerate the learning
efficacy of higher education students. With her expertise she has delivered a session
more than 150 on various titles. Her research areas include machine learning, deep
learning and Internet of Things, educational technology, educational data mining, vir-
tual education, and educational psychology. She holds a PhD degree in Information
and Communication Engineering from Anna University, Chennai, India. She has
published 28 articles in various journals and conference proceedings and contributed
chapters to the books. She has won two best paper awards: one in IEEE and another
in Springer conference. She has to her credit one book publication titled on “Theory
of Computation” and seven Indian patents provisionally published and waiting for
examinations.

Rajesh Kumar Dhanaraj is Professor at Symbiosis Institute of Computer Studies

and Research (SICSR), Symbiosis International (Deemed University), Pune, India.
He obtained his PhD degree in Information and Communication Engineering from
Anna University, Chennai, India. He has presented research papers at conferences,

vii
viii About the Editors

published articles and papers in various journals, and contributed chapters to books.
His research interests include Wireless Sensor Networks and Cloud Computing. He is
an Expert Advisory Panel Member of Texas Instruments Inc., USA.

Balamurugan Balusamy has served up to the position of Associate Professor in

his stint of 14 years of experience with VIT University, Vellore. He had completed
his bachelors, Master’s, and PhD degrees from top premier institutions in India. His
passion is teaching and adapts different design thinking principles while delivering
his lectures. He has published more than 30 books on various technologies and visited
over 15 countries for his technical course. He has several top-​notch conferences in his
resume and has published over 150 of quality journal, conference, and book chapters
combined. He serves in the advisory committee for several startups and forums and
does consultancy work for industry on Industrial IOT. He has given over 175 talks in
various events and symposium. He is currently working as Associate Dean, at Shiv
Nadar University, and teaches students and does research on Blockchain and IoT.

Sherimon P.C is presently working as a Faculty of Computer Studies, Arab Open

University, Oman. He has 22 years of international experience in administration,
teaching, and research in universities. He has completed numerous research projects
and consultancy projects too. His Google Scholar citation is 316, h-​index is 7, and
i10 index is 6.
Contributors
Prasanth A. Ahmed A. Elngar
Vel Tech Rangarajan Dr. Beni-​Suef University
Sagunthala R&D Institute of Egypt
Science and Technology, Chennai, India
Arthi K.
Chidhambararajan B. SRM Valliammai Engineering College
SRM Valliammai Engineering College Chennai, India
Chennai, India
Kannan K.
Sangeetha B. Annapoorana Engineering College,
AVS Engineering College Salem, Tamil Nadu, India
Salem, India
Meena K.
Wasana Boonsong GITAM University
Rajamangala University of Technology Bengaluru, India
Srivijaya
Thailand Thamarai Selvi K.
SRM Medical College Hospital and
Rajesh Kumar Dhanaraj Research Centre
Symbiosis International (Deemed Kattankulathur, India
University),
Pune, India Umapathy K.
Sri Chandrasekarendra Saraswathi
Lakshmi D Viswa Maha Vidyalaya (SCSVMV)
VIT Bhopal Deemed University, Kanchipuram,
India Tamil Nadu, India

MohanaPriya D. Veena K.
KIT-​Kalaignarkarunanidhi Institute of Sathyabama Institute of Science and
Technology Technology
Coimbatore, India Chennai, India

Rajalakshmi D. Yamuna K.S.

Sri Sairam Institute of Technology Sona College of Technology
Chennai, India Salem, India

Nguyen Tien Dung Fathima M.

Thai Nguyen University of Medicine Sri Krishna College of Engineering and
and Pharmacy Technology
Thai Nguyen, Vietnam Coimbatore, India

ix
newgenprepdf

x Contributors

Vetrivel M. Aravind Krishnaswamy Rangarajan

R P Sarathy Institute of Technology, Ghent University
Salem, Tamil Nadu, India Belgium

Senthamil Selvi M. Omkumar S.

Sri Ramakrishna Engineering College Sri Chandrasekarendra Saraswathi
Coimbatore, India Viswa Maha Vidyalaya (SCSVMV)
Deemed University, Kanchipuram,
Archana M.A. Tamil Nadu, India
Sri Chandrasekarendra Saraswathi
Viswa Maha Vidyalaya (SCSVMV) Shreya Kakkar
Deemed University, Kanchipuram, VIT Bhopal University
Tamil Nadu, India India

Deepa N. Devi T.
Saveetha University Saveetha University
Chennai, India Chennai, India

Mariya Ouaissa Dineshkumar T.

Moulay Ismail University Sri Chandrasekarendra Saraswathi
Meknes, Morocco Viswa Maha Vidyalaya (SCSVMV)
Deemed University, Kanchipuram,
Manjula P. Tamil Nadu, India
Saveetha Institute of Technical and
Medical Sciences, Kokilavani T.
Chennai, India Sri Krishna College of Engineering and
Technology
Mathivanan P. Coimbatore, India
KIT-​Kalaignarkarunanidhi Institute of
Technology Pham Chien Thang
Coimbatore, India TNU-​University of Sciences
Thai Nguyen, Vietnam
Deepa P.V.
Arunachala College of Engineering Ta Thi Nguyet Trang
for Women International School, Thai Nguyen
Kanyakumari, India University
Thai Nguyen, Vietnam
Kavitha P.V.
Sri Ramakrishna Engineering College Saranya V.
Coimbatore, India Saveetha University
Chennai, India
Janhvi Rajyaguru
VIT Bhopal University India Ranju Yadav
VIT Bhopal University
Hari Krishnan Ramachandran Bhopal, India
SASTRA Deemed to be University
Thanjavur, Tamil Nadu, India
 1 Advanced Enabling
Technologies of
IoMT in Personalized
Healthcare
K. Arthi* and B. Chidhambararajan
Department of Electronics and Communication Engineering,
SRM Valliammai Engineering College, Chennai, India
*Corresponding Author: K. Arthi ([email protected])

1.1 INTRODUCTION
The rapid increase in the aging population, the prevalence of chronic health issues,
and the rise in the spread of pandemic diseases pose a strict need for healthcare
services for people affected by economic conditions and reachability. The lack of
immediate attention to people affected by prolonged illness and accidents increases
the number of deaths every year. Informational communication between people and
doctors remains a major hindrance in such cases. To avoid such provocations, many
researchers started to keep on exploring new technologies to provide innovative
solutions.
The Internet of Medical Things (IoMT) has played a significant role in transforming
traditional healthcare services into digitized healthcare platforms to enhance the
quality of lives and improve patient outcomes [1]. IoMT is a medical network infra-
structure that has the potential to bring interaction between people, processes, things,
and networked objects connected through the Internet. IoMT ensures a promising
solution by providing remote monitoring of people’s health status and reporting to
hospitals, medical experts, caretakers, etc.
Before the evolution of the Internet of Things (IoT), patients’ visits to hospitals
were restricted and there was no way to provide continuous monitoring and diag-
nosis [2]. The IoT has transformed healthcare services, unleashing their potential
by connecting medical devices to the Internet to collect information from invalu-
able data. IoMT empowers physicians to promote continuous patient engagement
and remote monitoring of patients’ health, reducing the length of hospital stays at
reduced costs.
Due to the increase in demand for a 5G network, the design and production of
healthcare devices like smartwatches, fitness trackers, smart gadgets, and smartphones

DOI: 10.1201/9781003405450-1 1
2 Technological Advancement in Internet of Medical Things

are getting dramatically more attention. A survey based on Fortune Business Insights
predicts that the global IoMT market would expand from $30.79 billion to $187.60
billion between 2021 and 2028, with an average annual growth rate of 29.5%, down
from 71.3% in 2020. The convergence of communication technologies and its fusion
of advanced enabling technologies like artificial intelligence (AI) and blockchain
technology deliver 4P services, namely, predictive, preventive, personalized, and par-
ticipative, to shift traditional healthcare to a new era of digitalized healthcare [3].
Collaboration of emerging technologies facilitates remote monitoring, chronic dis-
ease management, electronic health records management, medication adherence,
and so on.

1.2 IOMT COMMUNICATION PROTOCOLS

Connectivity is necessary for data from the sensors to be exchanged through networks
to share information locally or centrally. The communication protocols employed in
the healthcare system are also based on its performance and functional requirements.
A communication protocol can be chosen based on use cases and distances, which
may be for short-​range or long-​range communications. For sending data gathered
from sensors to a smartphone nearby, short-​range protocols like ZigBee and Bluetooth
are recommended. The information from the patient’s wearable sensors can be sent to
the doctor or healthcare provider via SMS or the Internet via long-​range connectivity
such as Long-​Term Evolution (LTE). Wireless body area networks are utilized for
inter-​body communication. Table 1.1 shows the important communication protocols
widely used in IoMT applications [4].

1.3 FOG-​BASED ARCHITECTURE OF IOMT

IoMT is the concentrated embodiment of IoT technology in the medical field. In
general, components of the IoMT system architecture comprise medical devices,
medical applications, data management servers, and patient monitoring systems
connected through the Internet within the network. The author presented a fog–​cloud
architecture for COVID-​19 detection that comprises three layers, namely, user, fog,
and cloud layers [5]. An in-​depth knowledge of IoMT architecture is essential in
knowing the significance of the technological components involved and how the asso-
ciation between components of the medical ecosystem was established. As medical
applications are latency-​sensitive, a tri-​layer IoMT architecture was proposed to pro-
vide real-​time monitoring to save lives by reducing the latency and enhancing the
quality of healthcare services. Figure 1.1 depicts a fog-​based IoMT architecture.

1.3.1 Perception Layer
In IoMT architecture, the uppermost layer is referred to as the perception or physical
layer, or sensing layer. The main functions of the perception layer were to provide
data collection and provide data access by transporting the data to the next higher
layer [6]. The key elements of the perception layer include sensors, actuators, medical
Advanced Enabling Technologies of IoMT 3

TABLE 1.1
Communication Protocols for IoMT Applications

Protocols Features Definition Uses

IEEE 802.15.4 • Topology: star or peer-​peer standard • To obtain information
• Supports 16-​bit and 64-​bit from body area
addressing networks and hospital
• Very limited battery environments
consumption
Infrared • Works up to the line of sight Protocol • Used in
or within a few meters thermometers and
• Used for point-​of-​care devices cameras
RFID • Eliminates the need for an Technology • Autonomous RFID
external power source. tags to develop body
• Real-​time location tracking area healthcare
systems.
• Passive RFID tag
used for ambient
monitoring of the
patient’s environment
Near Field • Supports in the range of a few Technology • Used in ingestible or
Communication centimeters. implantable sensors
• Ensures fast and secure
communication between two
devices
Bluetooth/​BLE • Data transmission over short Technology • Fitness monitors
distances • Portable medical
• Security enhanced within the devices
limited range
• Low initialization time
Wi-​Fi • Provides long-​range Protocol • Wireless Body Area
communication. Standard Network
• Require high power • Wearable handheld
consumption devices
• Robust connection in indoor
environments
ZigBee • Low-​cost and low-​power Protocol • Noninvasive
protocol healthcare
• Transfer data up to 200
m range.
• Designed for personal area
networks
6LoWPAN • Media access control protocol Technology • Internet-​connected
• Low bandwidth routers
• Covers a range of about 10–​
100 m
(continued)
4 Technological Advancement in Internet of Medical Things

TABLE 1.1 (Continued)

Communication Protocols for IoMT Applications
Protocols Features Definition Uses
LoRaWAN • Supports long-​range Protocol • Health monitoring
communication of 10 km system
• Battery-​operated device and
supports low-​power
• Used for both public and
private networks
COAP • Application protocol Architecture • Health domain
• Lightweight mechanism of • Remote monitoring
COAP suits well for resource-​ system
constrained nodes and lossy
networks
MQTT • Supports lightweight Architecture • Facilitates connection
communication protocol between a user device
• Works well for poor and the cloud
connections.
• Constrained processing and
memory resources
HTTP • Used for communication Architecture • Wearable
between the cloud and the thermometer
doctor • Remote monitoring
systems

FIGURE 1.1 Fog-​based IoMT architecture.

Advanced Enabling Technologies of IoMT 5

devices, and processing elements to acquire the desired data from the physical world.
The main component that drives the physical layer is sensors which mainly com-
prise environmental, physiological sensors and motion sensors to sense the data
about the patient along with contextual awareness. The heterogeneous data captured
by different sensors will be processed through hardware development platforms or
single-​board computers like Arduino or Raspberry Pi. The healthcare experts can
have direct access to the generated data by employing local routers and servers in and
around the premises. The collected data are passed to the fog layer for further data
processing.

1.3.2 Fog Layer
The fog layer acts as a connecting layer located between the things layer and the
cloud layer. The main functions of the fog layer are to provide data aggregation,
distributed data processing, real-​time response, and secured storage of medical data.
Fog nodes comprise local servers, access points, routers, set-​top boxes, and local
and border gateways [6]. Fog nodes fill the gap between devices and the cloud as
they lie close to the users, so an immediate response was achieved. Communication
between fog-​to-​fog nodes and fog-​to-​cloud nodes is wired or wireless based on net-
work connectivity. Distributed architecture, mobility, low latency, and bandwidth
efficiency are additional benefits of fog computing. The fog layer reduced the compu-
tational workload of devices in the physical layer as it was resource constrained. Fog
nodes collect the data from sensors and devices to perform data pre-​processing and
decision-​making. It enables alerts and notifications in case of emergencies or abnor-
malities noted in patients’ data.

1.3.3 Cloud Layer
The cloud layer is considered the service layer that contains repositories, computa-
tional resources, and centralized cloud servers to store and process the data received
from the fog layer. The cloud presents a virtual dynamic storage platform to intercon-
nect devices and perform computations on the generated massive heterogeneous data
[7]. Other than data storage, the cloud provides data analytics and data visualization
services that present the data in a structured format.
Cloud technology acts as a promising solution for warehousing the enormous
data generated by healthcare services. cloud platforms are used to provide efficient
on-​demand availability of cloud services, computation, and storage resources. The
cloud-​based medical information interfaces help healthcare experts derive insights
and take action from the generated data.

1.4 CLASSIFICATION OF IOMT DEVICES

IoMT is a connected infrastructure made up of an amalgamation of sensors, medical
devices, computing technologies, and medical applications interconnected through
the Internet. IoMT devices could be wearable devices, blood glucose monitors,
6 Technological Advancement in Internet of Medical Things

kiosks, or implants for medical treatment. Based on the application used, IoMT is
classified as on-​body segments, in-​body segments, community segment devices, and
asset management.

1.4.1 On-​Body Segment Medical Devices

On-​body devices are used to assess physical activity or track the health metrics of
an individual day-​to-​day. These devices are mostly clinical-​based or consumer-​
grade wearable devices used in the form of wristbands, smart apparel, textiles, and
smartwatches for collecting vital stats through medical apps and smartphones [8].
The devices equipped with sensors are used to track fluctuations in vital parameters
and make clinical evaluations using the collected on-​body data.

1.4.1.1 Consumer-​Based Medical Devices

Consumer-​based medical devices equip users to track longitudinal information of a
person’s heart rate and rhythm, blood pressure, daily activity, etc. Increasing use of
consumer-​based medical devices skyrocketed once people started to monitor their
health after the emergence of the pandemic situations. The most frequently used on-​
body wearables include FitBit Flex, Garmin vívosmart 4, iWatch, Move ECG –​a
wearable ECG monitor, HeartGuide –​a blood pressure monitor, and bio-​sensors.

1.4.1.2 Clinical-​Grade Wearables
Clinical-​grade medical devices are devices approved by medical regulatory authorities
and referred by clinicians for enhancing quicker diagnosis and treatment. Examples
of clinical-​grade wearables include smart belts for detecting falls, and gait and for
alleviating chronic pain in sensory nerves.

1.4.1.3 Implantable
An artificial implantable device is a biomedical device implanted into a human
body through an in vitro or surgical procedure used to treat a specific disorder. The
most common examples of implantable medical devices are artificial joints, cardiac
pacemakers, cochlear implants, and left ventricular assist devices (LVADs).

1.4.2 In-​Home Medical Devices

In-​home devices comprise remote patient monitoring (RPM) devices, medical alert
systems, and telehealth platforms to carry out constant remote monitoring of patients
without unnecessary visits to hospitals. The use of in-​home devices helps the patients
take care of their health from home and restore their lives with a faster recovery time.
Medical alert systems comprise wearable devices synchronized with a medical call
center to provide emergency services to those in need. They are used mainly for the
elderly or physically challenged people with a lack of mobility. RPM devices facili-
tate the healthcare management of people with chronic disorders by allowing them
to monitor, analyze, and report critical indicators to healthcare experts [9]. The tele-
medicine application aims to deliver epidemiological surveillance through medical
Advanced Enabling Technologies of IoMT 7

conference consultation. Telemedicine is used to deliver virtual clinical services

to patients, such as diagnosis and monitoring, via the convergence of information
technology.

1.4.3 Community Segment Medical Devices

The community segment encompasses the medical devices used for mobility services,
AI-​based emergency response intelligence systems, computerized kiosks, point-​of-​
care devices, and pharmaceutical logistics dispersed in cities or over a range of geo-
graphic areas to enable mobile, emergency, and remote services. Mobility services
are used to track the vital parameters of patients in transit. Medical personnel utilize
emergency response intelligence systems to track vital parameters in ambulance ser-
vices or safety-​restricted zones in the event of emergencies outside the hospital’s
perimeter. Point-​of-​care devices are used to eliminate laboratory visits by conducting
screening procedures and tests from distant locations [10]. Kiosks are customizable
computerized self-​service structures embedded with touch-​screen displays used to get
services without staff interaction. It is used to reduce patient waiting time, schedule
appointments, and dispense medicines. IoMT devices in logistics help sense tempera-
ture, shock, vibration, and humidity for the effective delivery of medical supplies and
pharmaceuticals needed by hospitals.

1.4.4 In-​Hospital Segment Medical Devices

The in-​hospital segment of medical devices comprises medical devices that were
inside hospital premises to handle asset management, patient flow management,
environmental and energy monitoring, and personnel management. These devices
provide a vast array of functions in several operational zones as shown in Figure 1.2.

FIGURE 1.2 Functions of in-​hospital segment medical devices.

8 Technological Advancement in Internet of Medical Things

1.5 ENABLING TECHNOLOGIES
1.5.1 Fog Computing
The rapid development of healthcare led to a rise in the number of IoMT devices,
which resulted in a massive quantity of data generated from sensors and medical
devices. Cloud computing offers data computation resources, processing facilities,
and data storage for the massive amounts of data generated by IoT devices. Though
it acts as a data warehouse, it still suffers from severe shortcomings like a lack of
immediate response, high data transmission costs, and security issues. Health data
processing demands low latency, real-​time tracking, and data heterogeneity issues.
Cloud computing fails to handle latency-​sensitive and real-​time applications in min-
imum response time as its data transmission back and forth across multiple hops
makes it unfit for applications like vehicular networks, healthcare, etc.
Fog computing was introduced as an alternative approach to cloud computing
that delivers data services effectively in both spatial and temporal dimensions [11].
Fog computing is a technological paradigm located close to the end users to pro-
vide instantaneous data analysis and results within milliseconds or seconds. It is a
decentralized architecture of computing between devices and the cloud that facilitates
data transmission with improved quality of service in terms of mobility, low latency,
bandwidth conservation, data privacy, and security [12].The author proposed a loss-
less Electroencephalography (EEG) data compression technique enabled by edge-​
fog architecture that aims to predict epileptic seizures [13]. The data reduction from
edge to fog and fog to cloud was obtained by a unique hybrid approach of KCHE,
i.e., a combination of k-​means clustering and Huffman encoding. The author used
Naive Bayes algorithm to predict epileptic seizures of patients at the fog gateway.
The proposed method attains an accuracy range of 99.53% to 99.99%. Machine
learning (ML) algorithms were employed to detect disease and determine diagnosis
options from clinical data analysis. The author presented IoMT-​based patient-​centric
monitoring for the prediction of heart disease [14]. ML classification algorithms were
used in the fog layer for quick analysis and prediction of heart disease, with improved
performance compared with other models.
In a remote health monitoring system, efficient and secure data transmission and
aggregation were presented to protect the data against security threats and privacy
issues from distant locations in a secure manner [15]. The proposed system comprises
(i) data encryption and decryption techniques; (ii) sensor, mobile, and fog nodes; and
(iii) a single cloud server. Data compression and lightweight symmetric key-​based
encryption are used for secured data communication. The secure message aggrega-
tion (SMA) algorithm at the mobile node is used to enhance aggregation efficiency.
The secure message decryption (SMD) algorithm at the fog node is used to enhance
data security at the fog nodes.

1.5.2 Blockchain Technology
Blockchain technology has a profound impact on the medical field in informa-
tion security management, digitalized tracking, health record management, and
Advanced Enabling Technologies of IoMT 9

accelerating clinical health flow [16]. Blockchain technology is a public digital

immutable ledger technology distributed across peer-​to-​peer networks that prevent
the healthcare system from being hacked by improving security and transparency
within the blockchain ecosystem [17]. A block is a container that comprises data, the
hash of the previous block, and the hash of the block to be linked together to form
a blockchain. The hash contains the timestamp of the new block created and trans-
actional information. Each time the patient visits the doctor, results or records are
recorded as a new block on the ledger that is appended to the preceding block to form
a chain that cannot be modified later.
Blockchain facilitates patient health records and drug supplies with improved
accessibility and accountability at the appropriate time without any data manipulation.
Blockchain cryptographic hash algorithms convert the records into a unique pattern
of codes, or hash function, that can be accessed only by the person who owns the data
[18]. Blockchain helps medical experts track the results and alert the caretakers in
case of emergencies by keeping health records. Blockchain technology documents all
kinds of transactions to provide a good deal of accountability, versatility, flexibility,
and immutability without any falsification in clinical trials.
Integration of blockchain technology in medical devices poses some problems in
terms of processing, storage, mobility, and bandwidth consumption as they are highly
resource-​constrained in terms of power, memory, and computational resources. An
ML-​based blockchain task scheduling method, is to protect data against fraud and
threats at various levels, including fog nodes and remote servers [19]. The perform-
ance was evaluated both in terms of hard and soft constraints for different medical
applications. PSLA2P, an authentication key agreement protocol, was proposed for
the IoMT architecture-​ deployed in-​ hospital and emergency conditions to estab-
lish communication between cloud servers and edge devices securely [20]. It also
allows legitimate users to access the data from cloud servers with low computational
overheads and better functionality. The author presented a fog-​enabled automated
patient monitoring scheme that enables remote monitoring in fog to detect critical
anomalies [21]. In addition, optimization of the fog nodes with blockchains facilitates
the secured exchange of information among diverse entities at reduced latency and
computational costs.

1.5.3 Artificial Intelligence
AI is steadily revolutionizing multiple domains, like smart voice assistants, self-​
driving cars, robots, and healthcare. The amalgamation of AI and IoT steers the IoT
systems into smart devices by incorporating intelligence algorithms into systems and
converting the data into actionable insights to improve the quality of human lives.
IoMT sheds light on many breakthroughs to transform traditional healthcare into a
smart healthcare service by delivering proactive care to patients with the new era of
modern medical applications like precision medicine, drug discovery, and manage-
ment, robotic surgeries, etc. AI algorithms were used in the IoMT system to secure the
data and alert healthcare experts and caretakers when there were any contradictions or
abnormalities in the marginal vital parameters. AI is used in point-​of-​care devices to
avoid the spreading of infectious diseases and track disease progression.
10 Technological Advancement in Internet of Medical Things

Machine learning and deep learning are subsets of AI used to derive insights and
make predictions from sensor-​generated data with improved speed and accuracy in
data analysis [22]. They are used in drug discovery, patient data analysis, medical
imaging, and disease anomaly detection. Integration of AI with blockchain technology
achieves a good level of trade-​off between energy and load by balancing load between
fog and cloud nodes using task scheduling. Personalized deep-​learning approaches
were used for the early diagnosis of diseases and to make clinical decisions.
An IoT-​based deep learning architecture was used to estimate the fetal Q wave,
R wave and S wave (QRS) complex in the abdominal Electrocardiography (ECG)
[23]. A deep learning model based on transfer learning allowed to run on the cloud
consists of three phases: (i) data processing and segmentation, (ii) time frequency and
representation, and (iii) training of the deep learning model and classification. A deep
learning model was used to predict if decompression surgery was recommended or if
it was curable by nonsurgical procedures [24]. The model performance was evaluated
by classical and generalizability evaluation, whereas logistic regression was used as a
benchmark model for results comparison.

1.5.4 Wearable Technology
Wearable devices are small wearables that comprise a wearable part and an elec-
tronic counterpart that can be incorporated into the clothing and accessories of the
users to keep track of vital health parameters. Nowadays, humans use wearables like
pedometers, smart watches, hearing aids, fitness trackers, and eyewear in their daily
lives. As noninvasive wearable devices are hands-​free, it simplifies repairing and
replacing them without any pain. Wearable technologies exist in forms such as sewing
thread, accessories, garments, and portables.
The wearables are also portable in nature. Examples include mobile phones,
cameras, and microphones. Compact, portable devices track and monitor fitness
measurements such as walking and running distances, counting steps, calculating
calories burned, measuring heart rate and pulse rate, and so on. Another form of wear-
able technology is in smart garment form that can be worn as a jacket and provides
feasible solutions to the unobtrusive measurement of new parameters. Wearables, in
the form of accessories, can supplement daily style with items such as smart jewel-
lery, smart headsets, or smart shoes. Monitoring and assistive technology are used
for chronic disease management, tracking Parkinson’s disease progression, and
monitoring bio-​signal waveforms.

1.6 ADVANCEMENTS OF IOMT
The proliferation of IoMT gained massive momentum in the field of healthcare to offer
benefits to people in the medical ecosystem in many ways. It has a lot of calibers to
redefine traditional healthcare services into modern digital healthcare using seamless
connectivity. IoMT made a huge step forward in recent years by offering smart ser-
vices to clinicians, patients, medical equipment, hospitals, and community segments.
For example, apart from the basic functions of a smartphone, it got improvised
concerning health by providing information about location, mobility metrics, track
Advanced Enabling Technologies of IoMT 11

FIGURE 1.3 Advancements in the IoMT.

workouts, heartbeat irregularity, calorie tracker, and heart rate notifications [25, 26].
Some of the unique advancements of IoMT that outshined the previous innovations
are explained next and outlined in Figure 1.3.

1.6.1 Health Monitors [B HEAD]

Most people are susceptible to diabetes due to sedentary lifestyles and poor eating
habits, which need recurrent monitoring to track glucose levels. FreeStyle Libre 2 is
a small, unobtrusive continuous glucose monitoring system used to monitor glucose
levels for adults and children. Walgreens TrueMetrix Bluetooth Blood Glucose Meter,
Eversense, and Dexcom G6 are some of the best glucose monitors for diabetes man-
agement. These devices activate alarms and alert the caretakers when the levels cross
the threshold limit.
Many blood pressure monitors are available on the market in the form of fitness
trackers, smart watches, wristbands, bracelets, etc. People like to use devices that
are conspicuous and unobtrusive in nature. In that scenario, Valencell Ear Pressure
Devices in the form of earbuds and Vivae Blood Pressure Ring took a huge market
to monitor blood pressure and vital signs. The University of California, San Diego
(UCSD) developed a stretchable epidermal patch for tracing and monitoring heart rate,
lactate, blood pressure, and metabolites of glucose. KEYAR is a wireless intrapartum
monitoring device used to monitor uterine contraction, fetal heart rate, and maternal
heart rate at the time of labor to avoid complications.

1.6.2 Seizure Management
Seizure management is necessary as it may cause the sudden death of a person at any
time, even while sleeping. Proper monitoring avoids it by using seizure alert devices
12 Technological Advancement in Internet of Medical Things

in the form of headband wearables, smart watches, wearable bracelets, mattresses, and
cameras. For seizure management, MedPage movement alarm, seizure alert camera-​
SAMi, EpiWatch App, Sleep-​Safe pillow, and Emfit MM sleep monitor are used.
A brain sentinel device is used to detect uncontrollable and convulsive seizures using
an EMG recording with nigh-​on perfect accuracy. An IoMT-​based epileptic seizure
predictor, ForeSeiz, is designed as a wearable headband for patients with generalized
epileptic seizures [27]. The proposed system consists of threeElectromyography
(EMG) electrodes, NodeMCU ESP8266, and Firebase as the cloud. It uses Enhanced
Convolutional Neural Network for classification along with the Artifact Removal
Technique to avoid artifact contamination in EEG recordings. The authors evaluated
the proposed design on 12 real-​time subjects for validation and showed the results
with 97% accuracy.

1.6.3 Wearable
iRhythm Technologies designed a wearable Patch that detects atrial fibrillation and
prevents stroke. Reebok CheckLight is a wearable skullcap used as a sports activity
indicator to monitor the impact of head injuries. A Holter monitor is a noninvasive
wearable device recommended by doctors to keep track of heart rate and rhythm for
patients with arrhythmia.

1.6.4 Personal Assistants
Voice-​controlled personal assistants like Amazon Echo and Google Home provide
elderly people with assisted living and retrieval of information from electronic health
records. AI-​enabled smart speakers can be employed in home and infirmary rooms
to follow proper medical adherence and make patients friendly by getting engaged.

1.6.5 Implantable
Smart pills taken orally travel through the body, capturing images that help to ease
colon cancer at its early stages. The pills send information to peripheral devices
through the Internet or Bluetooth. Imec developed a smart pill used to examine the
gastrointestinal system along with muscle contractions. Capsule endoscopies have
become the best alternative for endoscopies without sticking a tube down into the
body. Pillcam COLON eliminates the necessity of a complex colonoscopy. Proteus
Patch keeps track of medication adherence to enhance the diagnosis rate. ABILIFY
MYCITE is a combination medicine of an antipsychotic tablet for the treatment of
schizophrenia or depressive disorders in adults with an ingestible sensor to record and
transmit time-​logged data.

1.6.6 Robots
Due to the recent advancements in medical technology, medical Robots are used in
many medical applications like robot-​assisted surgeries, prosthetics, robotic nurses,
clearing infections in clinical areas, and rehabilitation. ReWalk Personal 6.0 is
Advanced Enabling Technologies of IoMT 13

developed for personal use by those who have paralyzed or poorly handed arms or
hands. Surgical robot daVinci, a multi-​armed wonderbot, assists surgeons in the oper-
ating rooms by accessing magnetic resonance imaging (MRI) data of similar cases.
Xenex Germ-​Zapping Robot, an automated robot, is trained to kill infectious bacteria
in hospitals using ultraviolet (UV) rays. The PARO Therapeutic Robot is used to
reduce stress for elderly patients suffering from dementia and mood disorders.

1.6.7 Prosthetics
The tremendous growth of wearable technology and mobile platforms has put for-
ward its step in the field of prosthetics into reality by making suitable limb and skin
replacements in patients. TracPatch, a wearable device, Orthyo App, and BMPpathway,
an RPM system, are used in treating orthopedic patients after surgery and sharing the
data to improve recovery progress. Knee prosthetics with data transmission enable
clinician and patient contact to increase the mobility of patients in real time.

1.6.8 Hygiene Management
Implementation of proper hygiene practices among medical personnel in hospitals
and organizations is essential to preventing the spread of infectious diseases. IoT-​
based hygiene management systems are used to provide automatic alerts to medical
personnel to sanitize their hands on time in areas that demand higher standards of
hand hygiene.

1.7 CONCLUSION
In the Industrial 4.0 era, IoMT is on its way to transforming traditional medical ser-
vices into automated digital health services in every walk of life around us. Different
emerging technologies go hand in hand with IoMT to promote medical amenities
like telehealth services, sending reports, personalized care, and notification alerts.
Different medical requirements have opened the door to diverse types of medical
devices according to the needs of the well-​being of people.
In summary, this chapter developed the architecture of the IoMT that is suitable
for healthcare applications with low latency. A summary of IoMT communication
protocols and their application areas were discussed. IoMT was thoroughly reviewed
in the chapter with examples of healthcare applications. It also looked at some of the
essential enabling technologies that will help IoMT to reach new heights in the future.
The chapter ends by highlighting some recent and significant developments in IoMT
in the healthcare industry.

REFERENCES
1. Wagan, S. A., Koo, J., Siddiqui, I. F., Attique, M., Internet of medical things and
trending converged technologies: A comprehensive review on real-​time applications.
Journal of King Saud University–​Computer and Information Sciences, 34, 9228–​
9251, 2022.
14 Technological Advancement in Internet of Medical Things

2. Kumar, D. R., Krishna, T. A., Wahi, A., Health monitoring framework for in time rec-
ognition of pulmonary embolism using Internet of Things. Journal of Computational
and Theoretical Nanoscience, 15(5), 1598–​1602, 2018.
3. Firouzi, F., Jiang, S., Chakrabarty, K., Farahani, B., Fusion of IoT, AI, edge–​fog–​cloud,
and blockchain: Challenges, solutions, and a case study in healthcare and medicine.
IEEE Internet of Things Journal, 10, 3686–​3705, 2022.
4. Aruchamy, P., Gnanaselvi, S., Sowndarya, D., An artificial intelligence approach for
energy-​aware intrusion detection and secure routing in internet of things-​enabled
wireless sensor networks. Concurrency and Computation: Practice and Experience,
1–​33, 2023.
5. Akram, K. M., Sihem, S., Okba, K., Harous, S., IoMT-​fog-​cloud based architecture for
Covid-​19 detection. Biomedical Signal Processing and Control, 76, 103715, 2022.
6. Cheikhrouhou, O., Mershad, K., Jamil, F., Mahmud, R., A lightweight blockchain
and fog-​enabled secure remote patient monitoring system. Internet of Things, 22,
100691, 2023.
7. Jayachitra, S., Prasanth, A., Hariprasath, S., AI enabled internet of medical things
in smart healthcare. AI Models for Blockchain-​Based Intelligent Networks in IoT
Systems: Concepts, Methodologies, Tools, and Applications, 6, 141–​161, 2023.
8. Dao, N., Internet of wearable things: Advancements and benefits from 6G technolo-
gies. Future Generation Computer Systems, 138, 172–​184, 2022.
9. Boikanyo, K., Zungeru, A. M., Sigweni, B., Yahya, A., Remote patient monitoring
systems: Applications, architecture, and challenges. Scientific African, 20,
e01638, 2023.
10. Aljabr, A., Kumar, K., Design and implementation of Internet of Medical Things
(IoMT) using artificial intelligent for mobile-​healthcare. Measurement: Sensors, 24,
100499, 2022.
11. Elhadad, A., Alanazi, F., Taloba, A. I., Abozeid, A., Fog computing service in the
healthcare monitoring system for managing the real-​time notification. Journal of
Healthcare Engineering, 5337733: 1–​19 2022.
12. Wang, X., Wu, Y., Fog-​assisted internet of medical things for smart healthcare. IEEE
Transactions on Consumer Electronics, 1: 391–​399 2023.
13. Idrees, A. K., Idrees, S. K., Couturier, R., Ali-​Yahiya, T., An edge-​fog computing-​
enabled lossless EEG data compression with epileptic seizure detection in IoMT
networks. IEEE Internet of Things Journal, 9, 13327–​13337, 2022.
14. Chakraborty, C., Kishor, A., Real-​time cloud-​based patient-​centric monitoring using
computational health systems. IEEE Transactions on Computational Social Systems,
9, 1613–​1623, 2022.
15. Azeem, M., Ning, H., Ashraf, H., Zaman, N., Humayun, M., Fog-​oriented secure and
lightweight data aggregation in IoMT. IEEE Access, 9, 111072–​111082, 2021.
16. Bhaskar, K.B., Prasanth, A., Saranya, P, An energy-​efficient blockchain approach
for secure communication in IoT-​enabled electric vehicles. International Journal of
Communication Systems, 35, 1–​30, 2022.
17. Taherdoost, H., Blockchain-​based Internet of Medical Things. Applied Sciences, 13,
1287, 2023.
18. Andew, J., Isravel, D. P., Sagayam, K. M., Bhushan, B., Sei, Y., Blockchain for
healthcare systems: Architecture, security challenges, trends and future directions.
Journal of Network and Computer Applications, 215, 103633, 2023.
19. Lakhan, A., Mohammed, M. A., Nedoma, J., Martinek, R., Federated-​learning based
privacy preservation and fraud-​enabled blockchain IoMT system for healthcare. IEEE
Journal of Biomedical and Health Informatics, 27, 664–​672, 2023.
Advanced Enabling Technologies of IoMT 15

20. Abdussami, M., Amin, R., Vollala, S., Provably secured lightweight authenticated key
agreement protocol for modern health industry. Ad Hoc Networks, 141, 103094, 2023.
21. Cheikhrouhou, O., Mershad, K., Jamil, F., Mahmud., A., Lightweight blockchain
and fog-​enabled secure remote patient monitoring system. Internet of Things, 22,
100691, 2023.
22. Kavitha, M., Roobini, S., Prasanth, A., Sujaritha, M., Systematic view and impact
of artificial intelligence in smart healthcare systems, principles, challenges and
applications. Machine Learning and Artificial Intelligence in Healthcare Systems, 21,
25–​56, 2023.
23. Krupa, A. J. D., Dhanalakshmi, S., Lai, K. W., Tan, Y., An IoMT enabled deep learning
framework for automatic detection of fetal QRS: A solution to remote prenatal care.
Journal of King Saud University–​Computer and Information Sciences, 34, 7200–​
7211, 2022.
24. Jujjavarapu, C., Suri, P., Pejaver, V., Friedly, J. L., Predicting decompression surgery
by applying multimodal deep learning to patients structured and unstructured health
data. BMC Medical Informatics and Decision Making, 23, 1–​24. 2023.
25. Bhatt, M. W., Sharma, S., An IoMT-​based approach for real-​time monitoring using
wearable neuro-​sensors. Journal of Healthcare Engineering, 2023, 1–​10, 2023.
26. Subhan, F., Mirza, A., Su’ud, M. B. M., Alam, M. M., AI-​enabled wearable medical
internet of things in healthcare system: A survey. Applied Sciences, 13, 1394, 2023.
27. Prathaban, B. P., Balasubramanian, R., Kalpana, R., ForeSeiz: An IoMT based head-
band for Real-​time epileptic seizure forecasting. Expert Systems with Applications,
188, 116083, 2022.
 2 Deep Learning
Interpretation
of Biomedical Data
in IoMT
M. Senthamil Selvi,1* P.V. Kavitha,2 P.V. Deepa,3
and Ahmed A. Elngar4
1
Department of Information Technology, Sri Ramakrishna
Engineering College, Coimbatore, India
2
Department of Artificial Intelligence and Data Science,
Sri Ramakrishna Engineering College, Coimbatore, India
(Email: [email protected])
3
Department of Electronics and Communication
Engineering, Arunachala College of Engineering for Women,
Kanyakumari, India (Email: [email protected])
4
Faculty of Computers and Artificial Intelligence, Beni-​Suef
University, Egypt (Email: elngar_​[email protected])
* Corresponding Author: M. Senthamil Selvi
(Email: [email protected])

2.1 INTRODUCTION
To live a joyful and prosperous life, every individual in the universe deserves good
health. According to the World Health Organization, health is “a state of physical
and mental well-​being free of disease and disability.” Healthcare is the practice of
upholding and assessing one’s health through disease and illness prevention, treatment,
and improvement of well-​being and life quality. Medical errors are common in trad-
itional healthcare, posing a severe public health risk and threatening patient safety.
In today’s healthcare system, faulty diagnosis, prescription errors, improper
treatment, inadequate and insecure clinical facilities, or physicians’ lack of adequate
training can all lead to failure in emergency situations. These medical concerns are
prevalent in the majority of countries. To address the challenges that the medical
domain faces, IoMT plays a major role in saving patients’ lives through accurate
disease detection. IoMT replaces the traditional medical system and acts as a crucial
element in creating the upcoming healthcare industry.

16 DOI: 10.1201/9781003405450-2
Deep Learning Interpretation of Biomedical Data in IoMT 17

2.1.1 IoMT
IoMT is a networked environment composed of software packages and medical
equipment that interact with patients and healthcare professionals over the Internet
[1]. IoMT devices allow doctors to care for patients remotely, saving patients from
having to visit them in person. It connects patients with their doctors and allows med-
ical data to be transferred over a secure network, reducing the number of unnecessary
hospital visits. After the COVID-​19 outbreak, IoMT has been a hot topic in the rising
healthcare and telemedicine market. The Internet of Things (IoT) is sometimes known
as “Smart Healthcare.” The IoT has improved the way we engage with doctors [2].

2.1.2 IoMT Architecture
IoMT devices connect patients, physicians, hospital equipment, and wearable tech-
nologies to send health records over a secure network. IoMT is a smart healthcare
system made up of five components [3]. They are:

• Sensors Unit: This unit gathers biomedical signals from the patient.
• Processing Unit: This unit is used to process biomedical signals.
• Network Unit: This unit is used to send biomedical data via the Internet.
• Data Storage Unit: This is a device that temporarily or permanently stores data.
• Visualization Unit: Based on medical data, this unit uses artificial intelligence
(AI) algorithms to make decisions about diagnosis and therapy.

The components of the IoMT architecture are illustrated in Figure 2.1.

2.1.3 IoMT Technology
IoMT is also known as “Connecting Medical Devices to Networks.” It is built upon
three technologies. They are [4] discussed next.

Sensor Technology: The biosensor senses the patient’s health factors. It is of two
types: biological medical sensors and nonbiological medical sensors.

 Blood, respiratory, tissue, and other body parts are detected by a biological
medical sensor.
 Body temperature, mobility, electrical activity of the cardiac and muscles, as
well as other patient factors are all measured using a nonbiological medical
sensor.

FIGURE 2.1 IoMT Architecture.

18 Technological Advancement in Internet of Medical Things

Internet of Things Connectivity: IoMT data is transmitted directly and saved in a

repository (i.e., database). Clinicians can access the data over the Internet, allowing
them to provide better care to their patients.

Artificial Intelligence: Data is intelligently managed by IoMT devices [5]. It makes

intelligent decisions by removing irrelevant data from the medical system and
forecasting medical diagnoses using complicated analysis algorithms. This allows
medical practitioners to focus on patients instead.

2.1.4 Advantages of IoMT
The benefits of IoMT are listed as follows:

• Accessibility:
• It is the ability to access patient health data by doctors more quickly and
accurately.
• Low Per-​Patient Cost:
• It reduces costs by taking treatments at home and avoids unnecessary hos-
pital visits.
• Fast Per-​Patient:
• IoMT devices are faster and easy to implement.
• Improved Efficiency:
• Waiting time and visiting time to look into patients have been reduced, which
in turn improves efficiency because IoMT captures data from anywhere.

2.1.5 Challenges of Implementing IoMT

The following challenges are faced during the implementation of IoMT:

High-​Cost Infrastructure:
The cost of setting up IoMT infrastructure is very large.

Vulnerabilities in IoMT Security:

Securing millions of patient medical records is a challenge in maintaining the quality
of care.

Strain on Existing Networks:

Existing hospital networks have to be ready with additional support for connecting
devices and an information technology backbone that makes hospital networks fast
and secure. This is an ongoing issue because most hospitals are generally slow to
equip themselves with expensive IoMT devices.

2.1.6 Applications of IoMT in Healthcare

IoMT serves as a smart remote monitoring system that detects early signs of disease
and provides patients with suitable treatment. Some of the healthcare applications
Deep Learning Interpretation of Biomedical Data in IoMT 19

of IoMT [6] are prognostication of genetic disorders, anticipating the evolution of

the cardiovascular disease, irregular heartbeats, bone disorder, tumor prediction, and
early diagnosis of contagious ailments like dengue fever, coronavirus, malaria, etc.
It is used for identifying bone loss, precise peptides, and initial periodontal diseases,
detecting teeth areas affected by clenching and grinding, handling asthma issues,
measuring blood oxygen saturation, heart rate, and pulse measurements, sensing
dental problems, tracking blood glucose and levels of pressure, trying to identify
dental disease in the primitive phase, assisting in the pre-​diagnosis of diabetes, real-​
time tracking of viral and bacterial infections, and detecting tooth cavities. It is also
used to detect high body temperature associated with coronavirus infection, which is
useful for environment-​assisted living facilities, detecting adverse drug reactions, and
monitoring viral and bacterial infections in human body fluids such as sweat, urine,
and saliva in real time.

2.2 ROLE OF MACHINE LEARNING IN IOMT

IoMT’s AI technology plays a major role in detecting accurate early-​stage diagnoses
of diseases, which could save patients’ lives by allowing them to be cured. Machine
learning is frequently applied in the IoMT domain to assist the medical profession in
making accurate diagnoses, treating diseases, and providing high-​quality service to
patients. IoMT devices collect medical data from patients and transfer it to the cloud,
where a machine learning algorithm does the actual diagnosis. Machine learning uses
three important steps: pre-​processing, feature extraction, and classification for the
diagnosis process. The report is provided to medical practitioners after the diagnosis
process is completed.
In the healthcare industry, IoMT is currently quickly expanding due to the vast
volume of data. Owing to the processing and analysis volumes of patient health data,
machine learning methods in IoMT are taking a back seat in the healthcare industry.
Feature extraction is done manually throughout the diagnosis phase in machine
learning approaches. The workflow of machine learning is illustrated in Figure 2.2.

FIGURE 2.2 Machine learning.

20 Technological Advancement in Internet of Medical Things

2.3 DEEP LEARNING IN IOMT

2.3.1 Deep Learning Overview
Deep learning comes under the discipline of AI capable of dealing with extremely
complicated medical data and producing highly accurate diagnoses. [7]. It can manage
a vast amount of data and perform well. Without the assistance of a programmer, it
pulls the features from the data automatically. It necessitates more hardware support
in order to solve complex problems with high precision. It concentrates on high-​level
attributes rather than low-​level data attributes. It solves the issue from beginning to
end, rather than breaking it down into bits. It necessitates a longer execution time
because it incorporates a huge amount of constraints during the training phase. The
following layers are used to build deep learning algorithms:

1. Input layer: This is the initial layer, and it is where each node receives its data.
2. Output layer: This is the final layer, and it contains the problem’s output.
3. Hidden layer: Between the first and last levels, there lies a hidden layer. It
is made up of neurons that use an activation function to produce a response
after getting a collection of weighted inputs. When an image of a car is passed
through the initial layer (input) and middle layer (hidden), the final layer
(output) provides the result of a hatchback car or a sedan car.

Figure 2.3 illustrates the different layers of deep learning.

2.3.2 Deep Learning in IoMT

Deep learning plays a vital role in IoMT since it handles a massive volume of healthcare
records in the medical field. Deep learning’ inculcation into the IoMT domain keeps
an eye on enhancing the accuracy of disease diagnosis, medical image processing,
medical data analysis, and other related medical applications [8]. Figure 2.4 depicts
the structure of the deep learning IoMT framework.

FIGURE 2.3 Deep learning.

Deep Learning Interpretation of Biomedical Data in IoMT 21

FIGURE 2.4 Deep learning IoMT framework.

The deep learning IoMT framework comprises four layers [9]:

• Layer 1: Data collection from IoMT

• Layer 2: Cloud storage
• Layer 3: Deep learning mechanism
• Layer 4: End users

a. Data Collection from IoMT

This layer collects data from the human body’s wearable gadgets. Wearable devices
may consist of sensors and electronic circuits that monitor critical patient functions
such as blood pressure level, blood sugar level, oxygen level, breathing and pulse rate,
body temperature, urine report, and so on.

b. Cloud Storage
This layer sends health data from IoMT devices via a secure network to the cloud,
where it is kept for further analysis. Raw, unlabeled, and unfiltered data samples will
be collected.

c. Deep Learning Mechanism

The stages included in the deep learning mechanism are as follows: Pre-​processing
is the initial step in deep learning. The cloud data is preprocessed to eliminate noise
from the input, improve the required features, or resize the input. Upon preprocessing,
the data can be put into clever deep learning models, which extract features and con-
duct classification without the need for human intervention. The proposed system
detects early disease symptoms and delivers essential treatment to patients, thereby
saving their lives. It paves the way for more intricate health issues to be resolved in
22 Technological Advancement in Internet of Medical Things

emergency situations, as well as very accurate disease diagnoses for patients. The
next section discusses various deep learning models.

d. End User
Hospitals and medical practitioners could be the final users. In this layer, doctors have
access to patients’ medical records. The physician can notice the patient’s health dif-
ficulties by using a graphical user interface.

2.4 DEEP LEARNING ALGORITHMS IN IOMT

The various types of deep learning models employed in remote patient monitoring for
the prediction of extremely accurate diagnoses are as follows:

2.4.1 Deep Neural Network (DNN)

It involves input layer, intermediate layer, and output layer. The data is trans-
mitted between the processing layers via intermediate levels. The term “deep”
represents a higher number of hidden levels of processing. To deliver an efficient
output, the DNN collects the input data and processes it by using a linear or non-
linear connection. When the DNN receives the input, it calculates its probability
function for each layer’s probable output. The input is labeled based on the like-
lihood value.
DNNs can be used to address issues that have a lot of nonlinear relationships.
When the input data is represented as primitives in numerous layers, DNN uses add-
itional layers to composite aspects of the difficult problem. Deep architectures reveal
a plethora of new varieties not found in traditional neural designs. DNNs are feed-​
forward networks with only one direction of data flow (input layer to output layer).
To make a connection between virtual neurons, DNN first involves a series of virtual
neural connections and assigns weights to them at random. The random weights are
multiplied by those of the specified inputs to get an output between 0 and 1. If the
input data is not recognized by the network, an algorithm can automatically alter its
weight until it finds the correct mathematical mapping.

2.4.2 Restricted Boltzmann Networks (RBNs) or Auto Encoders

RBN [10] is a stochastic network used in deep learning. It is made up of two units
and there is no output unit. They are a visible layer and concealed/​hidden layer. Each
node of the visible layer is connected with its counterpart inside the concealed/​hidden
layer. The observation is aligned to the visible layer, while the feature extractor is
aligned to the hidden layer.
The RBN framework is depicted in Figure 2.5.
For learning the distributions of unlabeled training data, the RBN functions as a
generative model. It is a discriminative model that is used to classify labeled data.
They are called auto encoders, and they encode and decode data. The inputs are
encoded and converted into a list of numbers using a forward pass. A backward pass
Deep Learning Interpretation of Biomedical Data in IoMT 23

FIGURE 2.5 RBN.

decodes the inputs by taking these sets of integers to convert back into reconstructing
input data.

2.4.3 Deep Belief Networks (DBNs)

DBNs are feed-​forward neural networks made up of RBM models stacked one on
top of the other [11]. It solves the vanishing gradient problem. It consists of two
RBM models. The second visible layer of RBM is taken from the first hidden layer
of RBM model, then the output of the first RBM model can be used to train the
second RBM model. This approach is repeated until all of the network’s layers have
been trained. This network is more easily expandable. It consists of input and output
units. The visible layer is known as the input unit. The hidden layer is known as the
output unit.
DBNs allow for the extraction and classification of features. Layer-​by-​layer pre-​
training is used for feature extraction, back-​propagation is employed for classifica-
tion, and other tasks are tuned using a small labeled data set. The block diagram of
DBN is depicted in Figure 2.6.

2.4.4 Generative Adversarial Networks (GANs)

It is a kind of deep neural network made up of two networks [22]: generators
and discriminators. Adversarial training is another name for GAN. There are
two feedback loops in each network. Generator creates data whereas discrimin-
ator determines if the data generated is authorized. The steps of GAN process are
explained as follows:

• The generator network accepts random numbers as input and outputs

an image.
• The discriminator network receives the output of the generator, as well as
sample images from the original dataset.
24 Technological Advancement in Internet of Medical Things

FIGURE 2.6 DBN.

FIGURE 2.7 GAN.

• The discriminator outputs probabilities of 0 or 1, where 1 indicates true predic-

tion (i.e., authentic) and 0 indicates false prediction (i.e., fake). For example,
1 indicates a brain tumor prediction while 0 indicates a non-​tumor prediction.

The block diagram of GAN is depicted in Figure 2.7.

2.4.5 Recurrent Neural Networks (RNNs)

As in the intensive care unit (ICU), clinical medical data is often composed of a
multivariate time sequence of observations. The sensory data and lab test results are
recorded as part of the IoMT framework’s health monitoring of patients. The health
Deep Learning Interpretation of Biomedical Data in IoMT 25

FIGURE 2.8 RNN.

records have sporadic sampling, missing data, and differ in length. The common
models for learning from sequence data are Long Short-​Term Memory (LSTM) and
RNN [23].
RNNs deal with sequential data, which can flow in any direction. In an RNN,
all inputs and outputs are dependent on each other. It contains a memory that stores
the information from previous computations. The output depends on previous
computations to predict the next word sequence. The illustration of the RNN [12]
workflow is depicted in Figure 2.8.
The model parameters collected as output from various RNN model time steps
are shared. RNNs consist of a circuit that contains hidden to input and hidden to
output units.
RNNs resemble to the following equation:

H(t) = a(v + UH(t – 1) + WX(t)) (2.1)

O(t) = h + VH(t) (2.2)

where
• t stands for time.
• v indicates the offsets of the visible layer and h indicates the offsets of the
hidden layer.
• W denotes the weight joining the hidden and hidden units.
• V denotes the weight joining the hidden and output units.
• a represents the activation function.
• U represents the weights of hidden units at time t–​1 and t.
• X indicates the network’s input, whereas Y indicates the network’s output.
26 Technological Advancement in Internet of Medical Things

RNN is normally used in Natural Language Processing applications. Image

captioning applications employ the convolutional neural network (CNN)–​RNN
framework. The encoder uses CNN and the decoder uses RNN. The encoder
encodes the image automatically. From the encoded image, the decoder creates
word captions.

2.4.6 Long Short-​Term Memory

The output from the previous step is used as a source in the next phase in RNN [21].
It makes reliable forecasts based on recent data. RNN’s problems include:

• No provision for long-​term dependencies

• As the gap length grows, performance becomes inefficient

The aforementioned difficulties are dealt with by LSTM, which keeps the data for
a long time. It’s used for time-​series data processing, prediction, and classification.
The outcome of LSTM is determined by three factors:

1 Cell state: It comprises the network’s current long-​term memory.

2 Hidden state: It stores the output from the preceding time interval.
3 Input state: The current time step’s input data.

These three gates make up the LSTM module. They are.

Forget gate: This gate is in charge of eliminating unimportant data.

Input gate: This gate is in charge of adding data.
Output gate: This gate is in charge of transmitting data to the next hidden state.

The LSTM network is depicted in Figure 2.9. In LSTM, X(t) combined with pre-
vious output from cell h(t-​1) is passed to the input gate. “tanh layer” squashes the
input combination. This input combination is then sent to sigmoid-​activated nodes.
The squashed input is multiplied by sigmoid activated output and it is passed to the
forget gate.
In LSTM cells, state variables like S(t) and S(t–​1) are applied to input data to
reduce the risk of vanishing gradients and to assist the network in remembering or
forgetting the information. The output gate generates the output h(t) by multiplying
the tanh squashing function with a sigmoid-​activated node.

2.4.7 Convolutional Deep Neural Networks –​CNNs

CNN is a kind of neural network used to classify images [13]. It simplifies images
into a form that retains all of their attributes, making accurate prediction simple.
CNNs take multidimensional arrays as input data, such as two-​dimensional images
having three color channels. It consists of convolutional and pooling layers.
Deep Learning Interpretation of Biomedical Data in IoMT 27

FIGURE 2.9 LSTM architecture.

i. Convolutional Layer:
The input image is dealt with a filter (3 matrix) in this layer to produce a
convolved feature. The convolved features (Feature Maps) are transferred to
the next layer. The first layer gathers fundamental information like diagonal
or horizontal edges. The first layer’s output acts as input for the next layer.
As it progresses deeper into the network, it finds increasingly complicated
features.
ii. Pooling Layer
This layer shrinks the Convolved Feature’s spatial size. The data processing
time is sped up by reducing the dimensions. Down sampling is the term for
this method. On CNN, there are three different forms of pooling. They are
• Maximum/​Max Pooling
• Minimum/​Min Pooling
• Average/​Avg Pooling
• The highest value for every patch on that feature map is returned by Max
Pooling.
• The minimal value for every patch on that feature map is returned by Min
Pooling.
• Return the overall average for every patch on that feature map when using
Average Pooling.

However, Max Pooling outperforms both Min and Average Pooling in terms of per-
formance. The pooling layer generates a flattened two-​dimensional matrix, which is
sent into the Fully Connected layer for classifying the image.
The CNN architecture is depicted in Figure 2.10. The color blocks of the input
patch “A” are multiplied by the convolution kernel “K” to produce feature maps in
28 Technological Advancement in Internet of Medical Things

FIGURE 2.10 CNN architecture.

the convolution layer. The final output is generated using Max Pooling. Feature Maps
represent the matrix Aij and Kij. Convolutional matrix represents Cij. i and j indicate the
rows and columns of the matrix, respectively.

2.4.8 Transfer Learning Models

Transfer learning model trains a neural network for a specific problem and then uses
it to solve different problems by adding extra layers [14]. It reduces the time it takes
to train a neural network model and lowers the generalization error. The following are
common types of transfer learning models:

2.4.8.1 ResNet-​34
ResNet is an acronym for Residual CNN [24]. It overcomes the problem of vanishing
gradients. If the system is excessively lengthy, the gradients will be zero when the
loss function is computed. As a result, there is no learning and the weight is not
updated. The benefit of ResNet is that gradients can flow backward via later layers
to the starting filters. IoMT is utilized to detect the early phase diagnosis of leukemia
patients using the ResNet-​34 model. It is a CNN with 34 layers.
ResNet is an acronym for ResidualCNN. It overcomes the problem of vanishing
gradients. It is a CNN with 34 layers.
Deep Learning Interpretation of Biomedical Data in IoMT 29

FIGURE 2.11 ResNet-​34.

ResNet-​34’s architecture is depicted in Figure 2.11. VGG19 is the first layer. The
34-​layer plain network is the second layer, while the 34-​layer residual network is the
third layer. Other ResNet models, such as ResNet-​50 and ResNet-​101, are employed
in biomedical image categorization tasks.

2.4.8.2 DenseNet-​121
DenseNet is known as Densely Connected Convolutional Networks [15]. Deep con-
volutional networks’ depth is increased. Because every layer has its own weights to
learn, ResNet has a vast number of parameters. DenseNet layers, on the other hand,
only have a few parameters (e.g., 12 filters) and contribute a few additional feature-​
maps. DenseNet joins the layer’s output and incoming feature maps. ResNet sums
the layer’s output feature maps with the incoming feature maps instead of concaten-
ating them. DenseNet is divided into four categories: DenseNet121, DenseNet169,
DenseNet201, and DenseNet264. This chapter discusses DenseNet121 and
DenseNet201.
The DenseNet-​121 model can be used to diagnose COVID-​19 disease and to
identify the early-​stage detection of leukemia patients remotely using IoMT. The
following components make up DenseNet-​121:
30 Technological Advancement in Internet of Medical Things

FIGURE 2.12 DenseNet-​121.

• 1 –​7×7 convolution layer

• 58 –​3×3 convolution layer
• 61 –​1×1 convolution layer
• 4 –​AvgPool layer
• 1 –​Fully connected layer

Every dense block of DenseNet-​ 121 architecture contains two convolution

operations with various numbers of layers: a 1×1 sized bottleneck layer and a 3×3
kernel to conduct convolution operations. Figure 2.12 illustrates the diagram of
DenseNet-​121.
The layers of DenseNet-​121 are described as follows:

1. Convolution Layer: It consists of 64 7×7 filters and a 2 stride.

2. Pooling Layer: It uses Max Pooling of 3×3 and a 2 stride.
3. Dense Block 1: It consists of two Convolutions that are repeated sixfold.
4. Transition Layer1: It consists of one Convolutional layer and one Average
Pooling layer.
5. Dense Block 2: It consists of two Convolutions that are repeated 12-​fold.
6. Transition Layer 2: It consists of one Convolutional layer and one Average
Pooling layer.
7. Dense Block 3: It consists of two Convolutions that are repeated 24-​fold.
8. Transition Layer 3: It consists of one Convolutional layer and one Average
Pooling layer
9. Dense Block 4: It consists of two Convolutions that are repeated 16-​fold.
10. Classification: It uses Feature maps and Global Average Pooling layer for
classification.
11. Output Layer: It generates the output.

2.4.8.3 DenseNet-​201
The DenseNet-​201 model [16] is used to diagnose COVID-​19 disease, as well as
breast abnormalities and pneumonia. It consists of 201 layers and gives each layer
direct control over the actual input image and loss function gradients. It is a better
Deep Learning Interpretation of Biomedical Data in IoMT 31

FIGURE 2.13 DenseNet-​201.

FIGURE 2.14 VGG16.

option for image classification because it saves time and money. The DenseNet-​201
architecture is shown in Figure 2.13.
Because it employs fewer parameters and has a lower computational cost,
DenseNet gives better results than ResNet.

2.4.8.4 VGG16
VGG16 is a biomedical image classification model. It is a kind of CNN [17]. VGG16’s
architecture is shown in Figure 2.14.
VGG16 comprises three fully connected layers and five convolutional blocks.
Each block is made up of two or more convolutional layers as well as a Max Pool
layer. The RGB input image is 224 × 224 pixels in size. VGG16 is made up of two
layers: convolutional blocks and fully connected layers.

i. Convolutional Blocks
• Convolutional Block 1: It consists of two convolutional filters, each with 64
filters, and is based on Max Pooling.
• Convolutional Block 2: This block contains two convolutional filters. Max
Pooling uses 128 filters.
• Convolutional Block 3: This block contains three convolutional filters.
Max Pooling uses 256 filters.
32 Technological Advancement in Internet of Medical Things

• Convolutional Block 4: This block contains three convolutional filters.

Max Pooling uses 512 filters.
• Convolutional Block 5: This block contains three convolutional filters.
Max Pooling uses 512 filters.
ii. Fully Connected Layers
This layer is made up of three layers. A total of 4,096 hidden units and ReLU
activation make up the first and second layers. The final output layer contains
1,000 hidden units and Softmax activation.

2.4.8.5 InceptionResNetV2
CNN Inception ResNetV2 [18] incorporates the Inception and ResNet architectures.
The input image is 299 by 299 pixels, with a list of computed class probabilities as
the output. The inclusion of the ResNet network reduces the training time. Multiple
filters’ convolutional sizes are merged with a residual network. The layout of
InceptionResNetV2 is shown in Figure 2.15.
InceptionResNetV2 contains modified Inception-​Resnet blocks such as A, B, and
C blocks for extracting the features to classify the image.

2.4.8.6 U-​Net
The conventional CNN, which is used to handle biomedical image classification, has
developed into U-​Net [19]. U-​Net assists in locating and recognizing borders while
classifying each pixel. Both the input and output are of equal size. It has a U-​shaped
design [16]. It is divided into two sections. When it comes to biological applications,
U-​Net performs admirably.

• Contracting Path: The left section of the path is handled by the basic convo-
lutional process.
• Expansive Path: The right section of the path is used for segmentation
purposes.

Figure 2.16 shows the architecture of the U-​Net.

2.4.8.6.1 Contracting Path
Each step has two convolutional layers, and the number of channels ranges from
1 to 64 as the depth of the picture is increased via the convolution process. The

FIGURE 2.15 InceptionResNetV2.

Deep Learning Interpretation of Biomedical Data in IoMT 33

FIGURE 2.16 U-​Net architecture.

max pooling procedure minimizes the image size from 572×572 to 568×568 pixels.
The procedure is performed three more times. Without max pooling, two convolu-
tion layers are created once more. At this time, the image has been scaled to
28×28×1024.

2.4.8.6.2 Expansive Path
To get a precise prediction, the image upsizes to its original size in the expanding
path. The transposed convolution method expands the image’s size. The image is
resized from 28×28×1024 to 56×56×512 and then concatenated with the contracting
path image to create a 56×56×1024. This is followed by the addition of two more
convolutions, and the procedure is repeated three times more.
34 Technological Advancement in Internet of Medical Things

FIGURE 2.17 Fire Module–​SqueezeNet.

FIGURE 2.18 SqueezeNet architecture.

2.4.8.7 SqueezeNet
The essential element of this structure contains Squeeze unit and Expand unit, both
units together known as Fire Module [20]. The squeeze unit is made up of 1× 1 convo-
lution filters. The expanded unit is made up of a mixture of two convolutional filters,
1×1 and 3×3 convolution filters. It is a pre-​trained model for detecting pneumonia
and classifying it as bacterial or viral. The fire module and SqueezeNet structure are
depicted in Figures 2.17 and 2.18, respectively.

2.5 CONCLUSION
This chapter discussed the overview of IoMT in the medical industry. It highlighted
the IoMT architecture and the various components involved within the architecture.
It also provides an in-​depth understanding of IoMT technologies and the challenges
in the healthcare industry. Furthermore, it discussed the pros and cons of machine
learning models incorporation into IoMT in the medical field. It emphasizes the
benefits of inculcating deep learning models into IoMT. It explained how deep
Deep Learning Interpretation of Biomedical Data in IoMT 35

learning works in IoMT and discusses various deep learning algorithms used
in IoMT.
Deep learning models discussed in this chapter include DNN, RBN, DBN, GAN,
CNN, RNN, LSTM, and transfer learning. It emphasizes the significance of trans-
ferable learning paradigms. The transfer learning models described in this chapter
are the ResNet34, DenseNet121, DenseNet201, VGG16, InceptionResNetV2, U-​
Net, and SqueezeNet architectures. Finally, it goes over a few assessment metrics for
evaluating the performance of deep learning approaches. Transfer learning models of
deep learning into IoMT enhance the medical industry in data analysis, processing,
and accurate disease diagnosis.

REFERENCES
1. Dwivedi, R., Mehrotra, D., and Chandra, S., Potential of Internet of Medical Things
(IoMT) applications in building a smart healthcare system: A systematic review,
Journal of Oral Biology and Craniofacial Research, 12, 302–​318, 2022.
2. Ramasamy, M.D., Periasamy K., Krishnasamy, L., Dhanaraj R.K., Multi-​Disease
Classification Model using Strassen’s Half of Threshold (SHoT) Training Algorithm
in Healthcare Sector, IEEE Access, 9, 112624–​112636, 2021.
3. Al-​Dhaen, F., Hou, J., Rana, N.P., Weerakkody, V., Advancing the understanding of
the role of responsible AI in the continued use of IoMT in Healthcare, Information
Systems Frontiers–​A Journal of Research and Innovation, 1–​20, 2021.
4. Arvindhan, M., Rajeshkumar, D., Anupam Lakhan Pal, A., Review of Challenges and
Opportunities in Machine Learning for Healthcare, Exploratory Data Analytics for
Healthcare, 67–​84, 2021.
5. Kavitha, M., Roobini, S., Prasanth, A., Sujaritha, M., Systematic View and Impact
of Artificial Intelligence in Smart Healthcare Systems, Principles, Challenges and
Applications, Machine Learning and Artificial Intelligence in Healthcare Systems, 21,
25–​56, 2023.
6. Jayachitra, S., Prasanth, A., Hariprasath, S., AI Enabled Internet of Medical Things
in Smart Healthcare, AI Models for Blockchain-​Based Intelligent Networks in IoT
Systems: Concepts, Methodologies, Tools, and Applications, 141–​161, 2023.
7. Zemouri R., Zerhouni N., Racoceanu, D., Deep learning in the biomedical
applications: Recent and Future status, MDPI Applied Sciences, 9, 1–​40, 2019.
8. Greff, K., Srivastava, R.K., Koutník, J., Steunebrink, B.R., Schmidhuber J., LSTM: A
Search Space Odyssey, IEEE Transactions on Neural Networks and Learning Systems,
28, 2222–​2232, 2017.
9. Jayachitra, S., Prasanth, A., Multi-​ Feature Analysis for Automated Brain Stroke
Classification Using Weighted Gaussian Naïve Bayes Classifier, Journal of Circuits,
Systems and Computers, 30, 2150178:1–​23, 2021.
10. Yu, X., Zeng, N., Liu, S., Yu-​Dong, Z., Utilization of DenseNet201 for diagnosis
of breast abnormality, Machine Vision and Applications, Springer, 30, 1135–​
1144, 2019.
11. Huang, G., Liu, Z., Pleiss, G., Maaten, L.V., Kilian Weinberger, Q., Convolutional
Networks with Dense Connectivity, IEEE Transactions on Pattern Analysis and
Machine Intelligence, 1–​12, 2019.
12. Liu, F., Tan H., Song, D., Deep learning and its applications in biomedicine, Genomics
Proteomics Bioinformatics, 16, 17–​32, 2018.
36 Technological Advancement in Internet of Medical Things

13. Tawsifur, R., Muhammad Chowdhury, E. H., Amith, K., Khandaker Islam, R.,
Khandaker Islam, F., Transfer Learning with Deep Convolutional Neural Network
(CNN) for pneumonia detection Using Chest X-​Ray, Applied Sciences, 10, 1–​15, 2020.
14. Aayush, J., Neha, G., Dilbag, S., Vijay, K., Manjit, K., Classification of the COVID-​
19 infected patients using DenseNet201 based deep transfer learning, Journal of
Biomolecular Structure and Dynamics, 39, 5682–​5689, 2021.
15. Yang, Z., Xu, P., Yang, Y., Bao, B.K., A Densely connected network based on U-​Net
for Medical Image Segmentation, ACM Transactions on Multimedia Computing,
Communications, and Applications, 17, 1–​14, 2021.
16. Gao, H., Zhuang, L., Laurens Van Der, M., Densely connected convolutional networks,
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1–​6, 2017.
17. Sathish, R., and Ezhumalai, P., Intermodal sentiment analysis for images with text
captions using the VGGNET technique, ACM Transactions on Asian and Low-​
Resource Language Information Processing, 20,1–​14, 2021.
18. Nguyen, L. D., Lin, D., Lin, Z., Cao, J., Deep CNNs for microscopic image classifi-
cation by exploiting transfer learning and feature concatenation, IEEE International
Symposium on Circuits and Systems (ISCAS), 1–​7, 2018.
19. Siddique, N., Paheding, S., Elkin, C., Devabhaktuni, V., U-​Net and its variants for
Medical Image Segmentation–​A review of theory and applications, IEEE Access, 9,
82031–​82057, 2021.
20. Krizhevsky, A., Sutskever, I., Hinton, G.E., Imagenet classification with deep convo-
lutional neural networks, Advances in Neural Information Processing Systems(NIPS),
25, 1106–​1114, 2012.
21. Nandhini, P.S., Malliga, S., Monisha Sri, N., Jeevitha, K., LSTM: A Deep learning-​
based approach for the classification of intrusions in IoT based networks, International
Journal of Advanced Science and Technology, 29(9s), 26–​33, 2020.
22. Alankrita, A., Mamta, M., Gopi, B., Generative adversarial network: An overview
of theory and applications, International Journal of Information Management Data
Insights, Springer, 1, 1–​19, 2021.
23. Sherstinsky, A., Fundamentals of Recurrent Neural Network (RNN) and Long Short-​
Term Memory (LSTM) Network, Physica D: Nonlinear Phenomena Journal: Special
Issue on Machine Learning and Dynamical Systems, Elsevier, 404, 1–​16, 2020.
24. Sajja Tulasi, K., and Hemantha Kumar, K., Deep learning and transfer learning
approaches for image classification, International Journal of Recent Technology and
Engineering, 7, 427–​432, 2019.
 3 Machine Learning
for Decision Support
Systems in IoMT
T. Kokilavani,1* K. Kannan,2 K.S. Yamuna,3
B. Sangeetha,4 and M. Vetrivel,5
1
Department of Electrical and Electronics Engineering,
Srikrishna College of Engineering and Technology,
Coimbatore, Tamil Nadu, India
2
Department of Electrical and Electronics Engineering,
Annapoorana Engineering College, Salem, Tamil Nadu, India
3
Department of Electrical and Electronics Engineering,
Sona College of Technology, Salem, Tamil Nadu, India
4
Department of Electrical and Electronics Engineering,
AVS Engineering College, Salem, Tamil Nadu, India
5
Department of Electrical and Electronics Engineering,
R P Sarathy Institute of Technology, Salem, Tamil Nadu, India
*Corresponding Author: T. Kokilavani (Email: kokilavani@
skcet.ac.in)

3.1 INTRODUCTION
Hospitals utilize health prediction systems to monitor their patients and enable the
patients to get proposed treatment. Mostly, the healthcare services are mainly utilized
for the ambulance system during emergency events like natural disasters, etc. Some
of the hospitals lack real-​time data on the patient flow [1]. Under this situation, the
Internet of Things (IoT) establishes communication between virtual computers and
physical objects and provides real-​time data collection [2]. Healthcare is the process
of improving and maintaining people’s health through the diagnosis and prevention
of diseases.
The abnormalities in the body can be detected using diagnostic devices like
Computed tomography (CT) and Magnetic Resonance Imaging (MRI), etc. [3].
However, the modern healthcare systems have been affected by population increase.
Similarly, the demand for medical resources, such as nurses, physicians, and so on,
has also increased. It necessitates the development of standard healthcare systems
with high quality [4]. This can be achieved with the help of IoT. Doctors may readily
monitor patients’ information using healthcare monitoring schemes, which help them
to carry out an accurate diagnosis [5]. To transmit the patent data to the doctor, wireless
data transmission using handheld devices has been developed. While implementing
this, despite the advantages of the IoT, data security has to be maintained [6]. As a
DOI: 10.1201/9781003405450-3 37
38 Technological Advancement in Internet of Medical Things

result, a number of studies have concentrated on the utilization of IoT and machine
learning (ML) together to monitor medical issues and preserve data integrity.
In the health sector, IoT accompanied a new era, to communicate patients with
experts more proactively [7]. Hence, doctors can easily monitor patients’ informa-
tion through healthcare monitoring schemes. Various handheld gadgets have been
formulated to provide frequent wireless data transmission. IoT in conjunction with
ML evaluates emergency care requests and establishes a strategy for dealing with the
problem. In some hospitals, patients who need immediate medical attention have been
delayed because of improper control. This problem is more common in developing
countries, where hospitals are under-​equipped. In this situation, IoT with ML top-
ology assists hospitals to monitor the signs and symptoms of patients [8]. IoT sensors
monitor the health status of the patient and forecast illness trends and abnormalities.
This, in turn, will help patients to receive suggestions from different hospitals where
they can get treatment [9].
A medical monitoring platform with Zigbee Technology and BSN has been
developed for monitoring patients from the remote using sensors. To assess the
state of patients, modules like Zigbee protocol (IEEE 802.15.4) and sensors are
incorporated. The data is subsequently transferred through radio waves and visualized
via computers [10]. IoT uses sensor networks which are a combination of software
and hardware to enable patient monitoring and surveillance.
Devices such as the Raspberry Pi card, and blood pressure (BP), temperature,
and heart rate sensors fall under this category. Among these, Raspberry Pi cards
are companionable with cloud systems. Patient forecasting techniques are based
on ML and allow for fast sharing of patient data between hospitals. Historical data
of patients are utilized to estimate their future and help the hospital to predict their
health condition. While implementing ML, training classifiers can detect health
events using IoT devices. Clustering algorithms can identify and indicate the abnor-
malities of the patient to the healthcare providers [11]. It will be very useful for old
age people.

3.2 WORKING OF IOMT
The integration of IoMT into our daily lives has revolutionized healthcare, enab-
ling smarter and more connected medical systems. IoMT systems consist of several
interconnected layers that work together seamlessly, incorporating various devices,
sensors, and communication technologies (both wired and wireless). Figure 3.1
represents the structure and functionality of the layers involved in the IoMT layer.

3.2.1 Perception Layer
The perception layer in the IoMT continues to play a crucial role in gathering data
from a wide array of sources. These sources include smart objects, monitoring devices,
mobile apps, and various types of sensors like infrared (IR), medical based, smart
devices, radio frequency identification (RFID) tags, cameras, Global Positioning
System, etc. The primary function of the perception layer is to collect real-​world
Machine Learning for Decision Support Systems in IoMT 39

FIGURE 3.1 Different layers involved in IoMT framework.

information from these diverse sources and then digitally communicate it through a
reliable transmission infrastructure. This infrastructure can consist of either wired or
wireless networks, offering high-​performance indices for data transport.

3.2.2 Gateway Layer
The sensor must be linked to the gateway, as previously stated. Over networks,
gateways are set up to interact and store data locally or centrally. The gateways such
as RFID, WSN, cloud computing, etc., utilize different frequencies for communica-
tion [12–​15]. The devices like ZigBee, Bluetooth/​Ultra-​wideband (UWB)/​
Local Area Network (LAN), Ethernet, or WiFi connection can be used as the network.
Wide Area Networks (WAN) such as Global System for Mobile Communications
(GSM), which do not need connections but employ Wireless Sensor Network (WSN)
that can support back-​end servers/​applications and vast numbers of sensor nodes,
can be beneficial in some circumstances. Due to the stable connectivity of multiple
devices at the same time, 4G and 5G cellular networks are gaining popularity.
Huge communication potential has the ability to propel healthcare IoMT
applications forward and become a key driver. The many types of gates are listed
below. RFID is a technique for identifying a person. The short-​range communica-
tion gateway, with a range of 10 cm to 200 m, is equipped with a microprocessor
and an antenna. This gateway serves a dual purpose –​it can identify a specific object
and function as a reader that communicates with RFID tags using radio waves. The
information and data exchanged between the gateway and the tags are typically in the
form of an Electronic Product Code (EPC). RFID is unsecure even though the power
consumption is low. Therefore, it is not compatible with smartphones.
Random documents with unrelated
content Scribd suggests to you:
the Association and prevented from mingling with the true and sportsman like element
among school-boy athletes, the better will it be for athletics in this city.
The New Manhattan Athletic Club, or rather its athletic directors, were considerably
surprised, I know, at the attitude taken by this semi-professional element among the New
York school-boys. It had been their intention to offer a valuable trophy in the form of a
cup, to be contested for on this occasion, in addition to individual medals, and they had
even gone so far as to consult with the President of the National Association concerning
the order for this cup. But when they found that their interest in school-boy athletics was
apparently unappreciated, they gave up the idea entirely.
Fortunately the success or failure of the National Meet does not depend upon the entries
from the New York Association, and we may well rejoice if a lot of medal-hunters keep
away. Strong teams will come down from Maine, Massachusetts, and Connecticut, and
there will be representatives from New Jersey, and probably from other leagues, and the
sport will be good and clean, and the races will not be run with the sole idea of getting
money value in prizes at the end, but for the sake of the honor of winning on that day—
of the glory of sport for sport's sake.
At the recent Olympian Games the prizes were olive wreaths—plain, ordinary vegetable
growth; worth, say, ten cents a bushel, with perhaps fifty wreaths to the bushel. And yet
those dried branches brought home from Greece by the American winners are worth
more to them than any yellow metal they can get here. The young men who talk of
remaining away from the National meet, because the weight of the medals is not great
enough to suit their tastes, would do well to reflect on this: there is a greater object in
life than the collecting of medals.
The New England Interscholastic baseball season is practically closed, although there are
a number of games yet to be played. But Brookline has won the championship, having
played all its scheduled matches, and having won each of them. In the last game
Brookline defeated English High 6-0. Brookline played excellent ball both in the field and
at the bat, but E.H.-S. was weak all around. Some of the features of the game were
Nettleton's stop of Manning's hard hit in the fifth inning, Wise's clever throw from centre
to third in the eighth, putting out Cronin, and the heavy hatting of Lewis and Parker. A
review of the whole baseball season will be made in this Department as soon as space
enough becomes available.

C. S. D., Bayonne, N.J.—Any interscholastic association composed of at least two

schools may join the National Interscholastic Association upon applying for
membership. The field meeting this year will be on June 20, and is the first one ever
held by the Association, which was formed only last December.

The
Graduate.

ILL-TEMPERED BABIES
are not desirable in any home. Insufficient nourishment produces ill temper. Guard
against fretful children by feeding nutritious and digestible food. The Gail Borden Eagle
Brand Condensed Milk is the most successful of all infant foods.—[Adv.]
ADVERTISEMENTS.
A cream-of-tartar baking powder. Highest of all in leavening strength.—Latest United
States Government Food Report.
Royal Baking Powder Co., New York.
Arnold
 Constable & Co

Children's Wear.
Dimity Dresses,
Children's Guimpes,
Piqué Reefers, Mull Caps.

Misses' Outing Suits.

Misses' Shirt Waists.

Children's

Hand-Embroidered Underwear.

Broadway & 19th st.

NEW YORK.
EARN A BICYCLE!
We wish to introduce our Teas, Spices, and Baking Powder. Sell 75 lbs. to earn
a Bicycle; 50 lbs. for a Waltham Gold Watch and Chain; 25 lbs. for a Solid Silver
Watch and Chain; 10 lbs. for a beautiful Gold Ring; 50 lbs. for a Decorated
Dinner Set. Express prepaid if cash is sent with order. Send your full address
on postal for Catalogue and Order Blank.

W. G. BAKER, Springfield, Mass.

 Harper's Catalogue,
Thoroughly revised, classified, and indexed, will be sent by mail to any
address on receipt of ten cents.

This Department is conducted in the interest of Bicyclers, and the Editor

will be pleased to answer any question on the subject. Our maps and
tours contain many valuable data kindly supplied from the official maps
and road-books of the League of American Wheelmen. Recognizing the
value of the work being done by the L. A. W., the Editor will be pleased
to furnish subscribers with membership blanks and information so far as
possible.

Continuing the journey from New Haven to Springfield, which was left last
week at Hartford, the rider is advised to take what is called the East
Connecticut River Road: that is, leave Hartford by Main Street, and four
blocks from the City Hall—where, by-the-way, the United States Hotel gives
L. A. W. rates—turn into Morgan Street to the right, and run over the bridge
to East Hartford Street. On reaching East Hartford keep to the left, and take
the long road that runs never more than a mile away from the Connecticut
River. The road is in fairly good condition, and there are hardly any hills to
speak of during the whole run. The rider,
however, is of course advised to use side paths.
There is another route which may be taken
along the west side of the Connecticut, and
which is perhaps the better of the two. To take
this, run out Main Street direct instead of
turning right into Morgan Street, and keep on
until Windsor is reached. At the latter town
keep to the right and cross the Farmington
River, crossing the railroad, and running along
between it and the Connecticut River until
Windsor Locks is reached. This town is fourteen
miles from Hartford. At this point the
Connecticut should be crossed, and starting
from Warehouse Point, the rider should take the
road already described, running up the east
bank through Thompsonville towards
Springfield. Crossing the Massachusetts-
Connecticut line, he enters upon what is called
Long Meadow Street, runs into Long Meadow,
past Long Meadow station, and finally runs
close upon the Connecticut River again at
Pecowsic station. From Pecowsic the run into
Springfield to the Massassoit House is easily
found.
As has already been said in this Department,
this is not what may be called the Springfield
route from New York to Boston, and while the
stretch of country from Springfield to Worcester
is of course rideable, it is not a particularly Copyright, 1896, by
good road, and the country is not to any great Harper & Brothers.
extent picturesque, so that unless the trip is a
matter of making the journey—that is, if it is
simply for pleasure—the rider is advised rather to turn westward than
eastward, to ride a day or two in the Berkshire country, and then take a train
or trains for Worcester, continuing from Worcester to Boston on his wheel.
This trip from Worcester to Boston, and, in fact, from Springfield to Boston,
will be given in the near future in this Department to complete that particular
way of crossing Massachusetts from west to east.
Note.—Map of New York city asphalted streets in No. 809. Map of route
from New York to Tarrytown in No. 810. New York to Stamford,
Connecticut in No. 811. New York to Staten Island in No. 812. New
Jersey from Hoboken to Pine Brook in No. 813. Brooklyn in No. 814.
Brooklyn to Babylon in No. 815. Brooklyn to Northport in No. 816.
Tarrytown to Poughkeepsie in No. 817. Poughkeepsie to Hudson in No.
818. Hudson to Albany in No. 819. Tottenville to Trenton in No. 820.
Trenton to Philadelphia in No. 821. Philadelphia in No. 822. Philadelphia-
Wissahickon Route in No. 823. Philadelphia to West Chester in No. 824.
Philadelphia to Atlantic City—First Stage in No. 825; Second Stage in No.
826. Philadelphia to Vineland—First Stage in No. 827; Second Stage in
No. 828. New York to Boston—Second Stage in No. 829; Third Stage in
No. 830; Fourth Stage in No. 831; Fifth Stage in No. 832; Sixth Stage in
No. 833. Boston to Concord in No. 834. Boston in No. 835. Boston to
Gloucester in No. 836. Boston to Newburyport in No. 837. Boston to New
Bedford in No. 838. Boston to South Framingham in No. 839. Boston to
Nahant in No. 840. Boston to Lowell in No. 841. Boston to Nantasket
Beach in No. 842. Boston Circuit Ride in No. 843. Philadelphia to
Washington—First Stage in No. 844; Second Stage in No. 845; Third
Stage in No. 846; Fourth Stage in No. 847; Fifth Stage in No. 848. City of
Washington in No. 849. City of Albany in No 854; Albany to Fonda in No.
855; Fonda in Utica in No. 866; Utica to Syracuse in No. 857; Syracuse to
Lyons in No. 858; Lyons to Rochester in No. 859; Rochester to Batavia in
No. 860; Batavia to Buffalo in No. 861; Poughkeepsie to Newtown in No.
864; Newtown to Hartford in No. 865; New Haven to Hartford in No. 866.

Any questions in regard to photograph matters will be willingly answered

by the Editor of this column, and we should be glad to hear from any of
our club who can make helpful suggestions.
 SIMPLE CHEMISTRY FOR AMATEUR
PHOTOGRAPHERS.
In order to do good photographic work by method rather than by guess, it is
necessary to understand something of the nature of the chemicals used and
their effects. Even a slight knowledge of chemistry enables the amateur to
work understandingly and with far better results. We are therefore going to
give, for the benefit of our Camera Club, a few papers on chemistry as used
in photography, and shall try to make them so plain and simple that even the
youngest member will understand them.
Chemistry is that science which explains the composition of the substances
which compose the crust of the earth, the atmosphere which surrounds it,
and the water which occupies so much of the earth's surface. These
substances are called chemical elements. A chemical element is a simple
substance containing only one kind of matter, such as gold, silver, platinum,
hydrogen, oxygen, etc. According to the last report of Mr. F. W. Clark, the
chief chemist of the U.S. Geographical Survey, there are seventy-two known
elements. About thirty of these elements are used in the different processes
of photography.
Each element is represented by a symbol, this symbol being the first letter or
letters of the name of the element. The symbol of hydrogen is "H"; of oxygen
is "O"; of gold, "Au," the first two letters of the word "Aurum," the Latin name
for gold. Each symbol also stands for the weight of one of its atoms. (An
atom is supposed to be the smallest possible division of a substance.)
Hydrogen is the lightest element known, and is taken as the standard of
weight when comparing the weight of other atoms. The symbol "H" would
therefore not only stand for the element hydrogen, but for its weight, 1, or a
unit. An atom of oxygen is sixteen times as heavy as an atom of hydrogen,
and an atom of gold is 196 times as heavy.
In making up chemical compounds the chemical elements are combined in
different proportions, which, united, make a new substance. The way in which
these elements combine is always in the same proportion. The smallest
number of atoms which combine to form a new substance is called a
molecule. Take water, for instance, which is composed of hydrogen and
oxygen; it takes two atoms of hydrogen and one atom of oxygen to form a
molecule of water. These chemical combinations are expressed or written by
the symbols of the elements of which they are composed, called chemical
formulas. If two or more atoms of an element are used to form a chemical
compound, the number of atoms used is written directly after the symbol;
thus, H2O is the chemical formula for water.
Two well-known developing agents, pyrogallol—commonly called pyro—and
hydrochinon, are composed of the same chemical elements, carbon, oxygen,
and hydrogen, the only difference in their composition being that a molecule
of pyro contains one more atom of oxygen than the hydrochinon. The
chemical formula for pyro is C6H6O3, and the chemical formula for
hydrochinon is C6H6O2.
The chemical compounds employed in photography are used in the form of
solutions. A solution is the liquid combination of a liquid and a solid. A simple
solution is one in which the solid is entirely dissolved in the liquid, leaving the
liquid transparent. A saturated solution is a liquid containing as much of the
solid as can be dissolved in it and remain clear. In making saturated solutions,
unless the exact proportions are known, add the solid to the liquid until there
is a deposit of the solid at the bottom of the vessel containing the solution.
The clear liquid can then be turned off carefully into a bottle.
A solid dissolves much more quickly if it is first powdered. If one has no
mortar, put the solid inside a piece of muslin, lay it on a board or stone, and
pound with a hammer. When powdered, put the cloth and powder both into a
glass vessel, and turn the liquid over it. When the solid is dissolved, remove
the cloth. Another way in which to dissolve a solid more rapidly than by
mixing it with the liquid is to tie the powder in a cloth and suspend it in the
liquid.
In making up a formula for developing or toning, etc., be exact in the
measuring and weighing of the ingredients. Even a slight deviation from the
rule sometimes changes the action of the chemicals.
ADVERTISEMENTS.

"All is not
 Columbia
that Glitters."

Your pleasure and safety depend on knowing what is under enamel and nickel
before you buy a bicycle.
No question about Columbias. If you are able to pay $100 for a bicycle why
buy any but a Columbia?
See the Catalogue. Free if you call on the agent. By mail for two 3-cent
stamps

POPE MFG. CO.

HARTFORD, CONN.

Branch Houses and Agencies are almost everywhere. If Columbias are not
properly represented in your vicinity, let us know.

All Columbia Bicycles are fitted with

HARTFORD SINGLE-TUBE TIRES

UNLESS DUNLOP TIRES ARE ASKED FOR.

WE KNOW NO TIRES SO GOOD AS HARTFORDS.

WALTER BAKER & CO., limited.
Established Dorchester, Mass., 1780.
 Breakfast Cocoa

Always ask for Walter Baker & Co.'s

Breakfast Cocoa
Made at

DORCHESTER, MASS.

It bears their Trade Mark

"La Belle Chocolatiere" on every can.

Beware of Imitations.
There is
no
 Substitute
Every card of the famous DeLONG Hooks and Eyes has on the face and back
the words

See that

hump?

Richardson & DeLong Bros., Philadelphia.

Also makers of the CUPID Hair Pin.

Postage Stamps, &c.
 $117.50 WORTH OF STAMPS FREE
to agents selling stamps from my 50% approval sheets. Send at once for
circular and price-list giving full information.

C. W. Grevning, Morristown, N. J.

100 all dif. Venezuela, Bolivia, etc., only 10c., 200 all dif. Hayti, Hawaii, etc.,
only 50c. Ag'ts w't'd at 50% com. List FREE! C. A. Stegmann, 5941 Cote
Brilliante Ave., St. Louis, Mo

1000 Mixed Foreign Stamps, San Marino, etc., 25c.; 101 all dif., China, etc.,
10c.; 10 U.S. Revenues, 10c.; 20 U.S. Revenues, 25c. Ag'ts w'td at 50% com.
Monthly Bulletin free. Shaw Stamp & Coin Co., Jackson, Mich.

STAMPS! 100 all dif. Bermuda, etc. Only 10c. Ag'ts w'td at 50% com. List
free. L. Dover & Co., 1469 Hodiamont, St. Louis. Mo.

10 Rare Stamps free (postage 2c.). 5 Japan, all different, 8c. F. Jelke, 516
La Salle Ave., Chicago.
 The Fatal Letter.
Here is a letter and a puzzle all in one:
There was once a detective who had followed a criminal long and far, and
when he thought he had "run him to cover" he found the place empty and
only an open letter lying on the floor. He was overcome with chagrin, for it
had been said of him that "he had never failed to catch his man." In despair
he caught up the letter and read it. On the first reading it seemed a mere
succession of idle village gossip. He read it again, then sat down, and
pondered over the peculiar sentences all the night long.
His vigil, however, was not in vain, for three hours after dawn the criminal
was behind prison bars. Below is a copy of the "Fatal Letter." Can you
discover the secret message contained therein, the solution of which led to
the arrest of the criminal? Don't be discouraged because the detective spent a
night over it. Perhaps you are even sharper than the detective. He had no
clew. Neither can one be given to you. But this much may be said, the
message is not a haphazard affair, but follows a distinct plan.

THE LETTER.

Beloved Sister,—Dwellers of this town have been much excited over a

little affair of recent happening. A servant stole from a rich woman what
she called a ewer; it was brought from over the ocean. It was of fine
porcelain with heavy gilt edges and a calm summer scene painted on the
sides. A man of this town will soon begin sheep-raising. If that old ewe is
saleable send her on. It is the druggist who wants her; R. Jones is his
name. He will marry soon a girl of this city; his wedding gift is a diamond
necklace of elegant and chaste design. It must have cost every cent of a
thousand dollars.
The man who lived next door is dead. He took a draught of poison and
only lived two hours after. His wife was once the belle of the town. She
keeps crying, "I'll take poison myself." Yet he was a poor provider; they
had meat only once a month, and their table was always ill supplied. He
was also as meddlesome as a flea and of very uncertain temper. Quite
lately he quarrelled with me because of my statement concerning a lump
of iron ore which he owned. Answer this soon and don't forget about the
old ewe. Ah, another bit of news. The woman at the candy store has
found a dye that has turned her gray tresses as black as a coal.
 B
rother Tom.

How We Interested Our Chapter.

A few weeks ago the Allen Chapter, 715, was on the verge of "breaking
up." The writer, who is president and a Founder, thought long and at last
found a way by which he could start an interest again. It was this. We
began to publish a paper called the Allen Courier. Only one copy was
made, and that was written. A circuit was started—i.e., a member, after
keeping the paper a day, handed it to the next on the list. In this paper
the writer, who is editor-in-chief, inserted stories written by the members,
clippings, Chapter news, etc., and so started a fresh interest in the
glorious old Allen. At the next meeting all were on hand.

C
laude T. Reno.
A
llentown, Pa.

For the Natural History Club.

One day in July, in company with a friend, I crossed a lake near my home
in search of interesting specimens. By chance we came upon the nesting-
place of a colony of water turkeys. These birds are abundant here, but
this is the first time I ever found their nests. The latter appeared to be
several years old and were large structures, nearly flat on top. It was late
for eggs, and young birds were everywhere. They were covered with
white down, and presented a great contrast to the dark colors of the old
birds.
When the boat approached a tree containing young birds they would
tumble into the water, a distance of ten or more feet, where they would
dive long distances to escape us. The nests were in cypress-trees
growing in two or three feet of water. Sometimes as many as a dozen
 nests were in one small tree. Under these trees we shot two small
alligators. Perhaps the alligators knew the birds' habit of falling into the
water. We also found nests of the purple and Florida gallinule. I would
like to belong to a press association or corresponding Chapter.

E
d. H. Clute.
L
ake City, Florida.

Puzzle to Draw.
Two angles acute;
A triangle on rails;
Two little serpents
With twists in their tails;
Two spikes with a bar;
A tall headless tack;
Two angles acute
Which are placed back to
back;
A part of a circle
Two straight lines to
meet;
Two thirds of a cross;
A circle complete;
And lastly two angles. And do you
not find
A character loyal, brave, noble,
and kind?

W
ashington.

Knights in a Far-away Land.

The Table has two devoted members in distant South Africa. Their names are
George Uhlig and Ernest A. Chaplin. Writing under date of the middle of
February, they say the fruit season is just ended, and that apples are being
barrelled for winter, now coming on. Both attend Gill College, to which
students come from all parts of the colony, and their favorite games are
cricket and football; the former in summer and the latter in winter.
One of them remarks that from a perusal of the Round Table he thinks
baseball must be a good game, and that he would like to see a game—the
"New York's," for example. Both are fond of farm life, of hunting and fishing.
The principal birds are the dove, sparrow, fink, day-breaker, laughing, and
mouse birds. Both young men are stamp-collectors. Their address is Somerset
East, Cape Colony, South Africa.

Washington State Salmon.

In the State of Washington the fish industry comprises a good share of
the business. Salmon are the principal market fish, and are found in
abundance in the waters of Puget Sound and Gray Harbor. The salmon-
fishing season begins in September and closes the 1st of April.
In the first part of the season the "silver-side" salmon are alone caught,
and the run is very large. In the latter part the "steel-head" salmon is the
principal catch, the run being far less than in the former part. The
canneries only run during the period of time when the silver-sides are
running. Only Chinamen are employed in the canneries on Gray Harbor.
In the cannery at Cosmopolis eighty-five Chinamen are employed.
In the process of canning, the heads of the fish are first cut off, and the
salmon are dressed and washed until perfectly clean. They are next cut
into small pieces by what might be termed a "gang-chopper," after which
they are packed into cans. Every can has to be weighed. The salmon are
put up in one and two pound cans. The average sliver-side will weigh
thirteen pounds, for which the fishermen are paid thirteen cents apiece,
large or small, by the canneries.

R
euel M. Nims,
 C
osmopolis, Washington.

A Glimpse of Newfoundland.
The chief fisheries in which the public of St. John's are interested are the
cod, seal, salmon, and herring. These afford labor to the people of the
principal city of Newfoundland. Quite a few people are engaged in the
manufacture of the different kinds of gear used in taking fish, such as
lines, twines, nets, and cordage, also boats and tackle. The cod season
lasts longest. The seal fishery is the most valuable. Salmon and herring
are not much caught. The principal merchants of St. John's are engaged
in exporting fish. Times are very bad here.

B
. Bowering.
S
t. John's, Newfoundland.

This Department is conducted in the interest of Girls and Young Women,

and the Editor will be pleased to answer any question on the subject so
far as possible. Correspondents should address Editor.

The secret of being at ease wherever you are is a very simple one. It is only
this—Do not think about yourself. Bashfulness, awkwardness, and clumsiness
are caused by what we call self-consciousness, and as soon as we entirely
forget ourselves these pass away. A girl who writes to me complains that she
is so tall for her age that she cannot help being awkward. "The moment I
enter a room," she says, "I look about to see if any other girl is as tall as I
am, and I am always the tallest—a perfect bean-pole. Then I fancy that
everybody is sorry for me, and I cannot fix my attention on anything which is
going on. It makes me quite wretched. What shall I do?"
In the first place, my dear, your height, if you carry yourself well and hold
your head up, is a great advantage. Far from being a thing to regret, it is
something to be glad of.
Tall, or short, fat, and dumpy, or thin and pale, let the young girl never think
of this when she meets her friends. Instead, let her try her very best to make
the rest happy. If there is a girl in the room who is a stranger, or who seems
not to be having a pleasant time, single her out and entertain her. Your
hostess will be pleased with this sort of unobtrusive help, if it is kindly given.
A summer or two ago I happened to be paying a visit in a country house
where there were a half-dozen young guests. Among them were several
lovely girls from the South. I noticed that these girls had each some useful
social accomplishment. One played very sweetly, and she was always ready to
go to the piano and to play accompaniments for the violinist of the house
party, as well as to give us her dreamy nocturnes and slow sonorous marches
when we asked for them. Another sang, and she needed no urging when
there was a wish to hear songs. Still another played chess, and lent herself to
be partner to any one who wished that diversion. It was beautiful to watch
the sweet unconscious way in which these girls entertained the rest, never
putting themselves forward, but always to be depended on when it was a
question of how to pass an evening delightfully.
These are the days of out-door enjoyment, and my girls are playing golf and
tennis, and riding their wheels, and spending some portion of every day in
healthful exercise. Perhaps some of you like work out-of-doors as well as play,
and if there is a garden where you can dig and plant seeds and watch flowers
grow, or you have a poultry-yard with chickens and hens, or your talent for
the practical leads you to raise vegetables—radishes, pease, and lettuce
which grow for you will taste as no common market vegetables can. Keep in
the sunshine, girls. Sunshine means brightness and bloom for every one of
you.
 This Department is conducted in the interest of stamp and coin
collectors, and the Editor will be pleased to answer any question on these
subjects so far as possible. Correspondents should address Editor Stamp
Department.

One mail brought me two letters suggesting the formation of an exchange

society by the readers of the Round Table. Harold C. Day, Upland Farm,
Harrison, Westchester Co., N. Y., and Willis H. Kerr, Bellevue, Neb., both say
they would like to hear from any one interested. Other correspondents have
suggested the same thing at other times, and asked my opinion on the plan. I
regret to say it is not favorable. I have had some experience of exchange
societies, and have come to the conclusion that it is feasible only when some
capable man is at the head of the scheme who is willing to give his time and
experience to the plan, and that all sales are for cash only. All the larger
societies already have exchange circuits, and experience shows that common
stamps are not exchanged, and that valuable stamps must always be sent by
registered mail or by express, which is a considerable expense. The Dresden
International Society sends out books of stamps every year worth many
thousands of dollars; the leading society in New York has sent out five circuits
this year, aggregating about $2000 on each circuit. The first circuit was
completed a month ago. Stamps to the value of $1200 were taken. The
second circuit will be about the same. Almost all other societies have similar
plans.
Their method is quite simple. 1. All members who wish to contribute stamps
for exchange purchase a small blank book from the manager (Price 10c.). 2.
These books, filled with stamps, are sent to the manager, and when he has a
sufficient number they are done up in a package and sent out to the first
name on the circuit. 3. This person looks over the books, picks out what he
wants, sends a list of what he has taken to the manager, with P.O. money-
order for the amount. 4. He then sends the books to the second name on the
list, etc. 5. After the books have gone through the entire list the last man
returns them to the manager, who returns the unsold stamps to their owners,
and sends the cash (less commission) to those members whose stamps were
sold.
Some members buy very little and sell very much, others sell very little and
buy much. Before the books are sent out the manager examines them,
removes counterfeits, etc. Each man who takes out a stamp puts in its place a
"control" stamp with his number on it. These control stamps are bought of
the manager, and he only knows who has sold and who has bought.
It is expensive and troublesome. A much better plan is the old-fashioned one
of "swapping" stamps with one's comrades and friends.

H. B.—Your piece is a "Hard Money" token, not a coin. It has no money

value, but is very interesting.
L. K. Babcock.—See answer to H. B.
A. Ulmer.—The 6c. Hawaii, 1864 issue, is catalogued as worth 25c.

P
hilatus.

TO BICYCLISTS:

There is no better chain lubricant than Ivory Soap; it is a cleanly application

and perfect for this use.
Copyrighted, 1896, by The Procter & Gamble Co., Cin'ti.
The coolness is refreshing; the roots and herbs invigorating; the two together
animating. You get the right combination in HIRES Rootbeer.
Made only by The Charles E. Hires Co., Philadelphia.
A 25c. package makes 5 gallons. Sold everywhere.

THE

BALTIMOREAN PRINTING-PRESS
has earned more money for boys than all other presses in the market. Boys,
don't idle away your time when you can buy a self-inking printing-press, type,
and complete outfit for $5.00. Write for particulars, there is money in it for
you.

THE J. F. W. DORMAN CO.,

Baltimore, Md., U.S.A.

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookbell.com