California State University, San Bernardino California State University, San Bernardino
CSUSB ScholarWorks CSUSB ScholarWorks
Electronic Theses, Projects, and Dissertations O>ce of Graduate Studies
12-2021
INTEGRATION OF INTERNET OF THINGS AND HEALTH INTEGRATION OF INTERNET OF THINGS AND HEALTH
RECOMMENDER SYSTEMS RECOMMENDER SYSTEMS
Moonkyung Yang
Follow this and additional works at: https://scholarworks.lib.csusb.edu/etd
Part of the Arti=cial Intelligence and Robotics Commons, Business Analytics Commons, Business and
Corporate Communications Commons, Business Intelligence Commons, Computer and Systems
Architecture Commons, Databases and Information Systems Commons, Data Science Commons, Health
Information Technology Commons, and the Management Information Systems Commons
Recommended Citation Recommended Citation
Yang, Moonkyung, "INTEGRATION OF INTERNET OF THINGS AND HEALTH RECOMMENDER SYSTEMS"
(2021).
Electronic Theses, Projects, and Dissertations
. 1378.
https://scholarworks.lib.csusb.edu/etd/1378
This Project is brought to you for free and open access by the O>ce of Graduate Studies at CSUSB ScholarWorks.
It has been accepted for inclusion in Electronic Theses, Projects, and Dissertations by an authorized administrator
of CSUSB ScholarWorks. For more information, please contact [email protected].
INTEGRATION OF INTERNET OF THINGS AND
HEALTH RECOMMENDER SYSTEMS
A Project
Presented to the
Faculty of
California State University,
San Bernardino
In Partial Fulfillment
of the Requirements for the Degree
Master of Science
in
Information Systems and Technology:
Business Intelligence and Information Technology
by
Moonkyung Yang
December 2021
INTEGRATION OF INTERNET OF THINGS AND
HEALTH RECOMMENDER SYSTEMS
A Project
Presented to the
Faculty of
California State University,
San Bernardino
by
Moonkyung Yang
December 2021
Approved by:
Dr. Nasrin Mohabbati, Committee Chair, Information & Decision Sciences
Dr. Conrad Shayo, Committee Member, Information & Decision Sciences
Dr. Jay Varzandeh, Department Chair, Information & Decision Sciences
© 2021 Moonkyung Yang
iii
ABSTRACT
The Internet of Things (IoT) has become a part of our lives and has
provided many enhancements to day-to-day living. In this project, IoT in
healthcare is reviewed. IoT-based healthcare is utilized in remote health
monitoring, observing chronic diseases, individual fitness programs, helping the
elderly, and many other healthcare fields. There are three main architectures of
smart IoT healthcare: Three-Layer Architecture, Service-Oriented Based
Architecture (SoA), and The Middleware-Based IoT Architecture. Depending on
the required services, different IoT architecture are being used. In addition, IoT
healthcare services, IoT healthcare service enablers, IoT healthcare applications,
and IoT healthcare services focusing on Smartwatch are presented in this
research. Along with IoT in smart healthcare, Health Recommender Systems
integration with IoT is important. Main Recommender Systems including Content-
based filtering, Collaborative-based filtering, Knowledge-based filtering, and
Hybrid filtering with machine learning algorithms are described for the Health
Recommender Systems. In this study, a framework is presented for the IoT-
based Health Recommender Systems. Also, a case is investigated on how
different algorithms can be used for Recommender Systems and their accuracy
levels are presented. Such a framework can help with the health issues, for
example, risk of going to see the doctor during pandemic, taking quick actions in
any health emergencies, affordability of healthcare services, and enhancing the
personal lifestyle using recommendations in non-critical conditions. The
iv
proposed framework can necessitate further development of IoT-based Health
Recommender Systems so that people can mitigate their medical emergencies
and live a healthy life.
v
ACKNOWLEDGEMENTS
With all my heart, I would like to express my appreciation to Dr. Nasrin
Mohabbati and Dr. Conrad Shayo for encouraging me to develop my project. I
am also grateful to have my supportive family and friends who have been there
for me no matter what. There were ups and downs during the project timeline,
and I am proud of myself for getting through all the bitterness and sweet
moments. Along with this project, I am excited to go to the next chapter of my life
with more adventures and experiences.
v
TABLE OF CONTENTS
ABSTRACT .......................................................................................................... iii
ACKNOWLEDGEMENTS ..................................................................................... v
LIST OF TABLES ................................................................................................ vii
LIST OF FIGURES ............................................................................................. viii
CHAPTER ONE: INTRODUCTION ...................................................................... 1
Problem Statement .................................................................................... 2
Purpose ..................................................................................................... 5
Research Questions .................................................................................. 5
Structure .................................................................................................... 6
CHAPTER TWO: LITERATURE REVIEW ............................................................ 7
Data ........................................................................................................... 7
Artificial Intelligence (AI) ............................................................................ 8
Internet of Things (IoT) .............................................................................. 8
Internet of Medical Things (IoMT) .............................................................. 9
Mobile Health (mHealth) .......................................................................... 10
Wearable Devices .................................................................................... 13
Recommender System ............................................................................ 15
Health Recommender Systems ............................................................... 15
CHAPTER THREE: INTERNET OF THINGS IN SMART HEALTHCARE .......... 18
Architecture of Smart IoT Healthcare ....................................................... 18
Three-Layer Architecture .............................................................. 19
Service-Oriented Based Architecture (SoA) .................................. 21
vi
The Middleware-Based IoT Architecture ....................................... 22
IoT Healthcare Services and Applications ............................................... 23
A. IoT Healthcare Services ........................................................... 24
B. IoT Healthcare Service Enablers .............................................. 28
C. IoT Healthcare Applications ...................................................... 30
D. IoT Healthcare Services Focusing on Smartwatch ................... 38
CHAPTER FOUR: HEALTH RECOMMENDER SYSTEMS ............................... 43
Main Recommender Systems and Algorithms ......................................... 45
Content-Based Filtering ................................................................ 45
Collaborative-Based Filtering ........................................................ 46
Knowledge-Based Filtering ........................................................... 48
Hybrid Filtering .............................................................................. 49
Machine Learning Algorithms ........................................................ 51
CHAPTER FIVE: CASE STUDY AND DISCUSSION ......................................... 55
IoT-Based Health Recommender Systems Framework ........................... 55
Case Study .............................................................................................. 58
CHAPTER SIX: CONCLUSION AND FUTURE WORK ...................................... 65
Conclusion ............................................................................................... 65
Future Work ............................................................................................. 68
APPENDIX A: PYTHON CODE FOR CASE STUDY ......................................... 69
REFERENCES ................................................................................................... 74
vii
LIST OF TABLES
Table 1. mHealth Apps for general healthcare ................................................... 11
Table 2. Application of disease and examples of commercial wearable devices 14
Table 3. Most used technologies in the IoT field................................................. 20
Table 4. Comparison between Apple Watch and Samsung Galaxy Watch ........ 40
viii
LIST OF FIGURES
Figure 1. IoMT applications ................................................................................ 10
Figure 2. Most commonly used IoT architectures ............................................... 23
Figure 3. IoT-based Health Recommender Systems framework ........................ 56
Figure 4. Accuracy of different Machine Learning algorithms ............................. 60
Figure 5. Feature importance for heart diseases ................................................ 61
Figure 6. Correlation between the features of heart diseases ............................ 62
1
CHAPTER ONE:
INTRODUCTION
“The goal is to turn data into information, and information into insight.”-
Carly Fiorina
With the advent of the 4th industrial revolution and the development of 5G
technology, the Internet of Things (IoT) is increasingly being used in our daily
lives. For example, the importance of IoT in healthcare has recently risen due to
COVID-19 pandemic around the world. It is expected that people will receive
sufficient medical help and improve their quality of life due to the use of IoT in the
medical field (Islam et al., 2015). IoT-based healthcare systems are being used
in enhancing remote health monitoring, personal fitness programs, monitoring
chronic diseases, looking after the elderly people, and many other medical fields.
Embedded prediction and intelligence capacity in IoT devices enables
Remote Patient Monitoring (RPM), thus notifying patients about impending
emergencies before they turn into severe health problems (Ahmed & Kannan,
2021). According to a study conducted by Center of Connected Health Policy
“50% of the readmission rate is reduced by 30 days after using remote patient
monitoring (RPM) for heart failure patients” (Nasrullah, 2021). Another study
conducted by the Consumer Technology Association (CTA) showed that 49% of
patients had improved medical outcomes with employment of RPM and 42% of
patients took more ownership in health after using RPM (Pennic, 2019). In
2
addition, the Centers for Medicare and Medicaid Services has found out that
using RPM in Chronic Care Management programs decreased readmissions of
patients (Schurrer et al., 2017).
Remote Monitoring is a technology that assists in patient care. It can
collect and use a patient's biometric information such as blood pressure, glucose
levels, and oxygen saturation and then analyze the health data to transmit to
their healthcare providers (Hemanth et al., 2021). RPM accumulates patient’s
biometric data from wearable devices, patch-based sensors, Bluetooth biometric
devices, Smartphones, etc. Therefore, integration of IoT and RPM systems lets
users track their health condition. The ALTITUDE study discovered that 50% of
mortality rates in between one to five years decreased for patients who had
remote monitoring devices compared to patients who had in-person medical
checkup (Powell et al., 2013). Moreover, the outbreak of COVID-19 has led to
people paying more attention to RPM and IoT in healthcare, since it was hard to
see a doctor since the virus has contagiousness. Recently, RPM became
standard care to the patients who have implantable cardiac devices (Slotwiner et
al., 2015).
Problem Statement
According to the Centers for Disease Control and Prevention (CDC)
report, 41% of U.S. adults delayed or avoided healthcare treatment because of
concerns of COVID-19 in 2020, this included 12% of urgent care patients and
3
32% of routine care patients (Czeisler et al., 2020). Also, the CDC reported that
emergency visits for taking care of heart attack, stroke, and hyperglycemic crises
have declined since the beginning of the pandemic situation. This medical
situation would cause a more fatal crisis in public health. Avoidance of urgent
care and routine care could diminish a patient's chance to manage their health
conditions, and even further, the detection of diseases. Therefore, a contactless
healthcare system should be more actively used on a daily basis. Utilizing an AI
based healthcare system, IoT, and wearable devices could be the next chapter of
healthcare. It is critical for people to take care of themselves to prepare for
another situation similar to a pandemic. Throughout the COVID-19 pandemic, we
have been made aware of the importance of contactless healthcare treatment. It
is possible that we could encounter another pandemic which made meeting
others dangerous because of the contagious virus.
Also, some diseases require immediate health treatment, such as a
stroke, heart attack, and appendicitis. Anyone is susceptible to these time
sensitive diseases, and they can happen anytime. The most important thing to
prevent those diseases is to keep track of any symptoms and connect to a
medical center right away when the symptoms appear. Using medical IoT
sensors shows a patient's vital signs which are blood cholesterol, heart rate,
blood pressure, and other biological data depending on the patient’s installed
sensors (Akhbarifar et al., 2020).
4
In addition, the main problem of healthcare in the United States is
affordability. The study conducted by Kaiser Family Foundation in 2019
discovered that 29% of Americans decided not to take their medication at the
doctor's direction because of the cost of the prescription (Leonhardt, 2020). The
overall studies implied that we need to develop affordable devices to keep track
of personal health conditions. Therefore, it is important to study this project to
allow reduction of doctors visit costs and enhance the quality of healthcare
(Akmandor & Jha, 2017). Therefore, people could prepare for an emergent
medical situation and acquire knowledge about their health condition and that
treatment that is needed.
According to Gallup (2019), “25% of Americans report that they have
encountered medical treatment in a delay due to the medical cost” (Saad, 2021).
In addition, another study conducted by the American Cancer Society (2019)
found out that “56% of Americans have experienced medical financial difficulties”
(American Cancer Society, 2019). Those two studies implied that many
Americans cannot get medical treatments in time because of the medical cost.
The utilization of IoT can decrease the cost of medical treatment. In addition,
people can keep track of their health before severe diseases occur.
Considering all the above problems and challenges, this project will mainly
be focused on 1. how IoT would help to monitor personal health and 2. how IoT
integration with the Health Recommender Systems can enhance immediate
5
attention to the health conditions and affordability of healthcare by personal
recommendations.
Purpose
The purpose of this project is to present an overview of Data, Artificial
Intelligence (AI), Internet of Things (IoT), Internet of Medical Things (IoMT),
Mobile Health (mHealth), Wearable devices, Recommender System, and Health
Recommender Systems as well as how such technologies can play an important
role in fast and affordable healthcare services. Also, the objectives of the project
are to highlight how individuals can manage their health through IoT and how IoT
and Health Recommender Systems integration can provide more advanced
health treatment.
Research Questions
This study aims to show how modern technology can be utilized for
personalized health-related recommendations. Therefore, people can manage
their personal health on a daily basis before critical medical emergencies
happen. It also helps to take quick action in case of emergencies. Following are
the research questions that will be addressed in this study.
● What healthcare services can be provided by Internet of Things?
6
● What Internet of Things healthcare applications have been
developed for people?
● How do Health Recommender Systems work?
● What kind of machine learning algorithms are used in Health
Recommender Systems?
● How can recommender systems be used in the healthcare sector?
● How can integration of Internet of Things and Health Recommender
Systems improve responsiveness in identifying and treating health
issues?
Structure
This project is organized as follows: In Chapter 2, literature and past
studies in the related topics are reviewed. In Chapter 3, Internet of Things in
smart healthcare is discussed with architecture of smart IoT healthcare, IoT
healthcare services, and IoT healthcare applications. In Chapter 4, the Health
Recommender Systems using machine learning algorithms are reviewed. In
Chapter 5, a framework of IoT-based Health Recommender Systems is
suggested with a case study that is implemented by using Python to show
evaluating accuracy of machine learning algorithms, the important health
parameters of heart diseases, and correlation between the features of heart
diseases. Chapter 6 discusses conclusion remarks and the future work.
7
CHAPTER TWO:
LITERATURE REVIEW
To discuss the integration of Internet of Things (IoT) and Health
Recommender Systems, it is critical to know how previous works have been
done and set the starting point. Therefore, following concept represents: Data,
Artificial Intelligence (AI), Internet of Things (IoT), Mobile Health (mHealth),
Wearable devices, Recommender System, and Health Recommender Systems
for discussing the integration of IoT and Health Recommender Systems.
Data
Data in healthcare is generated in enormous quantities on a daily basis.
With that being said, the utilization of big data in healthcare can improve
healthcare treatment so the patients will decrease the medical impact on their
body. Generally, big data analysis consists of 6 V’s: volume, variety, velocity,
veracity, validity, and volatility (Jagadeeswari et al., 2018). Among the
characteristics, the three main features are volume, variety, and velocity. Volume
shows the quantity of the information to attain the respective goals. Variety refers
to the type of data that can be stored and analyzed. For example, variety in big
data can be videos, sounds, text, etc. Velocity tells the speed of when the big
data is generated or delivered to another (Kaur & Mann, 2017).
8
Artificial Intelligence (AI)
Artificial Intelligence (AI) is a technology which immerses human
intelligence so that computers have perception ability, learning ability, and an
ability to understand natural language through the computer programs (Kaur &
Mann, 2017). AI systems rely on their input data. AI in healthcare supports the
patient’s health monitoring with, for example, vital checks in real time (Kaur &
Mann, 2017). In addition, AI systems can explore the patients’ data, and then,
providing personalized health monitoring, recommendation, and treatment.
Internet of Things (IoT)
The Internet of Things (IoT) is a physical object that has a network
connection (Vermesan & Friess, 2013). There can be different types of devices,
such as medical instruments, home appliances, Smartwatch, industrial systems,
people, buildings, vehicles, and Smartphones. These devices are connected and
communicate with each other based on the required protocols to enable personal
online monitoring, process administration, tracing, and positioning (Vermesan &
Friess, 2014). IoT in healthcare can support health monitoring systems, wearable
health monitoring, remote health monitoring, Smartphone health monitoring, etc.
(Sahu et al., 2020). Patients keep monitoring their vital health conditions, for
instance, body temperature, blood pressure, blood glucose, respiration rate, and
pulse rate, using the sensor which is attached in the patient's body (Kumar &
9
Gandhi, 2018). Monitored patient’s vital health condition can be used for disease
prediction and further treatment (Madakam et al., 2015). Moreover, the monitored
data can be stored in the repository so that healthcare professionals have access
to the data for future medical treatments (YIN et al., 2016).
Internet of Medical Things (IoMT)
The Internet of Medical Things (IoMT) is the combination of Internet of
Things (IoT) with medical devices (Razdan & Sharma, 2021). IoMT is aimed to
manage patients’ health by using sensors implanted in medical objects and
transmitting the monitored data via network so that patients can communicate
with their healthcare providers (Vishnu et al., 2020). According to Figure 1, the
collected data from patients goes to the healthcare professionals, and then,
feedback goes back to the patients. In the near future, most medical devices
could connect and be monitored through the internet by healthcare professionals.
Such systems will reduce the cost of medical treatment and allow faster access
to medical care. In addition, IoMT with the integration of AI, big data, and cloud
computing will accelerate the IoMT usage in healthcare.
10
Figure 1. IoMT applications (Razdan & Sharma, 2021)
Mobile Health (mHealth)
Mobile Health (mHealth) is the utilization of mobile communication with
network technology for healthcare (Dutta et al., 2017). mHealth apps collect
user’s health information, nutrition, and wellness so that the apps can help
people with chronic diseases. In addition, users can track their workout schedule
11
and diet nutrition. Therefore, mHealth apps improve the user's overall health
condition by collecting user’s health related data. mHealth has been widely used
for communication between healthcare providers and patients and delivery for
healthcare services (Dutta et al., 2017). In Table 1 below, it shows what mHealth
apps are available and the description of apps.
Table 1. mHealth Apps for general healthcare
Apps
Health Assistant
Healthy Children
Google Fit
Calorie Counter
Water Your Body
Noom
Pedometer
Period Calendar
Period Tracker by GP
Apps
Instant Heart Rate
12
Cardiax Mobile ECG
ECG Self-Monitoring
ElektorCardioscope
Runtastic Heart Rate
Heart Rate Monitor
Kardia
Blood Pressure Watch
Blood Pressure
OnTrack Diabetes
Real Thermometer
Body Temperature
App
For Fever
Medisafe Meds
& Pill Reminder
Dosecast medication
Reminders
Rehabilitation Game
iOximeter
Eye Care Plus
SkinVision
AsthmaMD
Fight CF (cystic
fibrosis)
Hearing Test
uHear
Relax Noise 3
ReSound Tinnitus
Relief
BetterSleep
Sleep Cycle
13
FallSafety Home
-Personal Alert
Fall Detection
-Fall Alert Saves Lives
Calm
Headspace:
Meditation & Sleep
Daily Yoga
(Islam et al., 2015)
Wearable Devices
The definition of wearable devices is the devices that can be attached to
clothing and the human body with receptors and transducers (Xie et al., 2020).
Wearable devices can do patient monitoring, asset monitoring, tracking, early
medical interventions, and drug management (Banerjee et al., 2017). It can be
used in healthcare for cardiovascular diseases, Alzheimer, Parkinson’s disease
and other psychological diseases, asthma, obesity, and in-hospital monitoring.
Moreover, wearable devices not only support personalized health services but
also individualized portable devices and sensors (Guk et al., 2019). Portable
devices can be divided into wrists, body clothes, feet, heads, and body sensor
controlling devices (Kamišalić et al., 2018).
14
Table 2. Application of disease and examples of commercial wearable devices
Disease
Monitoring
Product
Category
Commercial Product
Cardiovascular
disease
Heart rate
Pulse rate
Wrist
Watch/band
HEM series of OMRON
Fitness
tracking
Heart rate,
Calories burned,
activity level,
Heart rate
variability, Body
temperature,
Cardiac electrical
activity (ECG)
Smart jewelry
Ear appliance
Watch/band
Ear-o-smart
Cosinuss’ One
Apple Watch
Samsung Galaxy
Watch
Cognitive
disorder
GPS
Wrist
Watch/band
Vega GPS bracelet
Sleep or stress
related disease
Heart rate
variability, Heart
rate
Wrist
Watch/band
Smart jewelry
Patch
Airo Health anxiety
tracker
Oura ring
Motiv ring
Go2Sleep
Kenzen Patch
Vital Scout
Metabolic
disorder
Glucose Hydration
Wrist
Watch/band
Ear appliance
Patch
GlucoWatch G2-
Biographer
GlucoTrack
Symphony
FreeStyle Libre
Dexcom Patches
LVL
Mosquito-borne
diseases
Temperature
Sweat patterns
Smart jewelry
TermoTell bracelet
Skin disease &
UV related
disease
Level of UV
Smart patch
Smart jewelry
MyUV Patch
Netatmo JUNE
Respiratory
diseases
Audio signal, heart
rate, accelerations
Cardiac electrical
activity (ECG)
Wrist
Watch/band
Smart patch
LG Watch Urbane
W150
Moto 360 2
nd
Generation
Savvy patch ECG
sensor
XYZlife Patch BC1
15
Skeletal
system
diseases
Movement
postural variation
gait
Smart shoes
CUR Smart Pain Relief
Valedo
(Guk et al., 2019)
Recommender System
Recommender System is a decision-making system used to filter
information depending on user’s preferences, interest, and previous activity
(Isinkaye et al., 2015). Therefore, it is helpful for users to make their choice by
giving out suitable options in the information overloaded world (Sahoo et al.,
2019). Recommender System can be divided into seven different systems, such
as content-based filtering, collaborative-based filtering, knowledge-based
filtering, hybrid filtering, context-aware based filtering, demographic-based
filtering, and social-based filtering (Ertuğrul & Elçi, 2019)
Health Recommender Systems
The Health Recommender System (HRS) is a system that applies the
Recommender System in healthcare. Health Recommender Systems provide
medical information that is related to the patient’s medical history (Archenaa &
Anita, 2017). Health Recommender Systems can provide patients personalized
guidance into Clinical Diagnosis Systems (CDS) and provide personal
16
recommendations, for example, diet recommendations, follow-up alerts, list of
diagnosis, preventative care alerts, etc (Archenaa & Anita, 2017). The Quantified
Self (QS) is a concept of self-tracking of personal health conditions in physical,
biological, behavioral, or environmental information (Erdeniz et al., 2019). In
addition, the QS system will include more recommendation systems to support
the users using wearable devices, mobile phones, biosensors, and cloud
services. In the basis of Quantified-Self (QS), Virtual Coach, Virtual Nurse, and
Virtual Sleep Regulator can be used for improving personal health conditions.
Virtual Coach helps to schedule a workout plan and Virtual Nurse supports a
physical activity plan depending on the user’s medical history and Virtual Sleep
Regulator assists insomnia users to improve their sleep quality by recommending
their sleep plan and physical activity. For example, collaborative filtering-based
recommender systems use K-Nearest Neighbors (KNN) approach to find
similarities among the population and recommend a topic that might be
interesting and helpful for the targeted users by high chances (Erdeniz et al.,
2019).
In this chapter, Data, Artificial Intelligence (AI), Internet of Things (IoT),
Internet of Medical Things (IoMT), Mobile Health (mHealth), Wearable devices,
Recommender System, and Health Recommender Systems have been
reviewed. The literature review helps to understand how various technologies
can come together to help monitoring personal health conditions. It also gives an
insight to develop an idea which is an integration of IoT and Health
17
Recommender Systems. Therefore, this study is focused on highlighting how can
integration of IoT and Health Recommender Systems help responsiveness and
affordability in healthcare.
18
CHAPTER THREE:
INTERNET OF THINGS IN SMART HEALTHCARE
The Internet of Things (IoT) is the physical object having a connection with
the network and embedded with technologies and sensors so that devices could
communicate with other tools and systems over the network. IoT makes
industries such as health, wearables, transportation, CCTV, manufacturing,
agriculture, smart cars, traveling, banking and smart homes develop their
potential (Nawara & Kashef, 2020). Using IoT guarantees industries to track,
monitor, communicate, and operate their remote devices for their users.
IoT could enhance its capability integrating with big data, Cloud
computing, and AI. AI is a broad concept of technology that aims to create
human intelligence in computer programs. Machine learning is a subset of AI that
tries to train machines/computer programs to learn (IBM Cloud Education, 2020).
This project studies the IoT integration with recommender systems as a machine
learning technique and a subset of AI.
Architecture of Smart IoT Healthcare
Current developments in computational power and server-based
computing have strengthened the use of AI systems in many industries (Maranda
et al., 2018). Such developments also have enabled reusability of AI tools with
connected devices and correlated sensors. There are three most commonly used
19
IoT architectures: Three-Layer architecture, Service-Oriented based Architecture
(SoA), and Middleware-based architecture (Lombardi et al., 2021). These
architectures are explained in detail below.
Three-Layer Architecture
Three-Layer Architecture consists of perception layer, network layer, and
application layer (Lombardi et al., 2021). Perception layer is the first phase of
interacting with environments and certain objects by collecting the data and
information (Lombardi et al., 2021). At this level, devices and sensors should be
able to communicate with the target so that they could exchange information.
Also, devices are equipped with computing abilities to use smart technologies,
self-identification, self-diagnosis, and self-testing (Abdmeziem et al., 2015). In
addition, there are special characteristics in smart objects, such as
communication, identification, addressability, sensing and actuation, embedded
information processing, localization, and user interface (Lombardi et al., 2021).
Communication is the essential aspect of the perception layer for updating data
and services and for attaining their goals. Identification is the key property to be
individually identified. Addressability makes the objects reachable to do a remote
control and configuration. Sensing and actuation is the part of the perception
layer responsible for collecting information and data from the surrounding
situation and utilizing it using sensors and actuators. Embedded information
processing is a critical feature for calculation functions to process the outcome of
devices and sensors. Localization is also an important feature to track the
20
physical location of devices and users. User interface gives a proper platform to
communicate with users. Table 3 shows most used technologies for operating
smart devices (Čolaković & Hadžialić, 2018).
Table 3. Most used technologies in the IoT field
Scheme
Used Technologies
Communication
Zigbee, Bluetooth, Wi-Fi, Near Field Communication (NFC),
Radio-Frequency IDentification (RFID), etc.
Identification
Electronic Product Code (EPC), Ubiquitous Code (uCode),
Quick Response (QR), etc.
Addressability
IPv4, IPv6
Sensing
e Actuation
Micro Electro-Mechanical Systems (MEMS) e Micro-Opto-
Electro-Mechanical Systems (MOEMS), embedded
sensors, etc.
Embedded
information
processing
Field Programmable Gate Array (FPGA), Programmable
Logic Controller (PLC), microcontrollers, Single-board
computer, System-on-Chip (SoC).
Localization
Global Position System (GPS), Galileo, etc.
User interface
Displays, remote control, etc.
(Čolaković & Hadžialić, 2018)
The Network layer conveys the information and data from the perception
level to the application layer. In this layer, it has all network technologies and
protocols to obtain a stable connection. There are different kinds of protocols
used in IoT devices and the protocols are decided depending on the given
21
situation such as the need for a certain transmission speed, the usage of each
node, and the network scale (Lombardi et al., 2021). Wired networks offer more
reliability and faster transmission (Sharma & Gondhi, 2018). Wireless sensor
networks (WSN) refer to a network configuration of sensor nodes and can
monitor the surrounding environment using sensors (Ghazal et al., 2021). WSN
makes it possible to install an inaccessible situation and need fewer human
resources. Furthermore, using wireless protocol eases the ability to add and
remove the nodes when needed.
The Application layer has essential software to provide a specific service
to users. This is the stage that utilizes all the information and data which is
saved, filtered, and processed from the previous layer with analytic software so
that the data can be used in real IoT applications, such as smart wearable. This
layer also is referred to as a middleware. There are many platforms that
implement IoT applications such as Amazon AWS, Xively, and Microsoft Azure
(Lombardi et al., 2021).
Service-Oriented Based Architecture (SoA)
Service-Oriented Based Architecture (SoA) is the one of the most
commonly used architecture styles that is able to connect various functional units
of applications using interfaces and protocols (Lombardi et al., 2021). SoA is
composed of the perception layer-the network layer-the service layer-the
application layer. The newly added service layer manages service discovery,
service interfaces, service management, and service composition.
22
The Middleware-Based IoT Architecture
The Middleware-Based IoT Architecture is also called a five-layer
architecture (Ngu et al., 2016). The Middleware-Based IoT architecture consists
of the perception layer-the network layer-the middleware layer-the application
layer-the business layer. The Middleware-Based IoT architecture style has a
strength in connecting between data, applications, and users. Particularly, the
middleware layer has features that can collect and filter the obtained information
from the previous layer and process the data discovery and give an access
control of connected devices for applications. There are four advantages of the
Middleware-Based IoT architecture. It supports a variety of applications, and it
also can be able to run on various platforms and operating systems. Also, it
dispenses computing and the communication service among networks,
applications, and devices. Additionally, it helps standardized protocols to perform
and it also provides standard interfaces while supporting portability and
standardized protocols to do interoperability. Lastly, with a Middleware-Based IoT
architecture, it provides a secure interface for the applications. Figure 2 shows
most commonly used IoT architectures (Lombardi et al., 2021).
23
Figure 2. Most commonly used IoT Architectures (Lombardi et al., 2021)
IoT Healthcare Services and Applications
IoT-based healthcare systems are expected to improve the remote health
monitoring, personal fitness program, monitor the chronic diseases, look after the
elderly people, and many other medical fields (Islam et al., 2015). In this section,
IoT healthcare is explained by dividing it into services, service enablers and
24
applications. We also highlighted the IoT healthcare services focusing on
Smartwatch.
A. IoT Healthcare Services
IoT-based healthcare services are aimed to reduce the cost of healthcare
services, enhance the quality of people’s life, and improve the user’s experience.
IoT healthcare services are categorized as Ambient assisted living (AAL), Mobile
Health (mHealth), Wearable device, Adverse drug reaction, Community
healthcare monitoring, Children health information, Cognitive computing, and
Blockchain (Islam et al., 2015). Those have been outlined below.
Ambient Assisted Living (AAL). Ambient assisted living is a IoT healthcare
service that is powered by artificial intelligence (Pradhan et al., 2021). It is utilized
for elderly people to live their life more comfortably and safely. This healthcare
service provides autonomy and assistance for elderly individuals if an emergency
happens. IoT-based AAL architecture systemically provides healthcare services
to elderly people and disabled people (Shahamabadi et al., 2013). The bottom
line of the technology to apply this architecture in AAL uses IPv6-based low-
power wireless personal area networks (6LoWPAN) for active communications,
Radio Frequency Identification (RFID) and Near-Field Communications (NFC) for
using passive communications. This technological architecture has been
enlarged by integrating algorithms to detect elderly individuals’ problems in
regard to medical care with a professional medical knowledge. Moreover, when
25
medical emergency situations occur in elderly people, emergent detectors keep
monitoring their chronic conditions and medical emergencies (Sandeepa et al.,
2020).
Mobile Health (mHealth). Mobile Health (mHealth) is based on the mobile
phone platform using network systems to communicate, compute, and medical
sensors for healthcare services. The patients are able to share their personal
health data using a network area with their healthcare provider. Through the
network, healthcare providers can access patient’s health data, diagnose
symptoms, and actively provide treatment (Pradhan et al., 2021). Diabetic
patients can monitor their glucose level using mHealth technology so that
patients can manage their hypoglycemia (Istepanian et al., 2017). mHealth real-
time monitoring technology can detect abnormal signals of heart activity and
notify the condition to patients (Chuquimarca et al., 2020).
Wearable Devices. Wearable devices are devices that humans can attach
to their body and wear it (Xie et al., 2020). Wearable devices are noninvasive
and have potential to develop in different ways for human use. In other words, it
could be developed by integrating with different sensors in wearable devices for
people’s healthcare, such as watches, shirts, shoes, and wristbands (Singh et al.,
2020). The sensors that input in the wearable devices accumulate all the health
data from the patients and it is uploaded on the specific databases. Some
wearable devices are able to connect to mobile applications through the network.
26
IoT-enabled health monitoring devices can provide remote health monitoring with
several embedded sensors in heartbeat, blood pressure, and body temperature
(Wan et al., 2018). In addition, Electrocardiogram (ECG) and Electromyography
(EMG) are also utilized in IoT-based wearable systems (Kelati et al., 2018).
Therefore, Wearable devices can be utilized as monitoring systems for patients’
chronic conditions, such as sleep apnea, Parkinson disease, obesity, post-
traumatic stress disorder (PTSD), asthma, panic disorder, cardiovascular
diseases, pulmonary conditions, and hypertension (Piwek et al., 2016).
Adverse Drug Reaction (ADR). Adverse Drug Reaction (ADR) is defined
as “unpleasant or injurious reaction to patients occurred from intaking of
medicinal product” (Coleman & Pontefract, 2016). The ADR is an intrinsically
generic response and might happen by taking a single dose of medication or
prolonged administration or composite result from combination of different kinds
of drugs. The IoT-based ADR system is using a unique barcode/NFC-enabled
device to recognize each drug at the individual patient’s terminal (Jara et al.,
2010). The drug’s compatibility with the patients results in the patient's personal
allergic history and electronic health record. In addition, the IoT-based
prescription Adverse Drug Event (prescADE) has been developed to help
decrease the ADE so that patients can improve their healthcare services (Nakhla
et al., 2018). Moreover, patch testing in ADR has been proposed. Patch testing
supports the detection of T-cell-mediated/non-immediate drug eruptions
(Gonçalo & Bruynzeel, 2020).
27
Community Healthcare Monitoring. Community Healthcare monitoring can
be defined as a concept that institutes a healthcare network covering a local
community (Islam et al., 2015).
This healthcare service originated from an IoT-
based network system around a residential area, private clinic, and city hospital.
The concatenated community healthcare monitoring services are considered as
a cooperative network system so that it has to meet collective technical
requirements for collaborative healthcare services. IoT-based healthcare
monitoring systems in a rural area found out to be more energy-efficient. In
addition, for a better cooperative network, it is necessary to incorporate
authentication and authorization mechanisms. The community healthcare
network is considered as a “virtual hospital” and a resident healthcare information
service platform is founded on a four-layer structure. The medical facilities and
the service platform are sharing data to obtain people’s health records and to
access remote medical advice (Wang et al., 2012).
Children Health Information. Children Health Information is aimed to let
children and their parents know about their children’s overall health condition
including children’s nutritional values, psychological condition, and behavior. One
study developed an IoT-based framework which can monitor a child's
psychological and physical state and also suggested an IoT-based healthcare
network where a medical device can connect with a mobile app (Sutjiredjeki et
al., 2020). The system requires individual physical parameters such as heart
28
rate, height, temperature, SpO2, and weight so this information can be analyzed
by the doctors or health professionals over the mobile app. In addition, it is
possible for parents and teachers to monitor children’s food habits using m-
health service (Briseno et al., 2012).
B. IoT Healthcare Service Enablers
In this subsection, the technologies that enhance analytics in healthcare
services have been reviewed such as cognitive computing and blockchain.
Cognitive Computing. Cognitive Computing is a system that can analyze
the problems in the human brain. Upon the development of artificial intelligence,
now IoT-based devices are integrated with the artificial intelligence and sensor
technology to imitate the algorithms of the human brain to solve the problem.
Moreover, Cognitive Computing plays a big role in analyzing patterns in a
massive amount of data (Behera et al., 2019). Cognitive Computing also
improved the capacity of a sensor to process healthcare data and adjust to the
environment. Utilizing the cognitive computing in the IoT system supports
healthcare providers to keep track of a patient's condition and give a better
treatment. Smart healthcare system based on the EEG which is using Cognitive
Computing can make a decision about the pathological state of the patient (Amin
et al., 2019). It also helps the patients who are in a medical emergency such as
stroke and heart attack. One study also has proposed a mechanism of
transmission of the cognitive data so that it helps to detect, record, and analyze
29
individual patient’s health data (Kumar et al., 2019). Furthermore, when it comes
to the emergent situation, Cognitive Computing transmitted the patient's health
data with priority.
Blockchain. Blockchain technology is employed to solve the data
fragmentation problems. It also supports the healthcare centers to build up a
connection in the data repositories (Satamraju & B, 2020). Data fragmentation is
a crucial point of secure data sharing (Pradhan et al., 2021). Data fragmentation
can cause information gaps between healthcare providers. Therefore, the use of
blockchain technology in the healthcare sector can solve the data fragmentation
problem and reduce the barriers that can result in delaying the treatment process
of patients. In addition, blockchain technology guarantees the security of sharing
sensitive medical information and improves transparency between the patients
and doctors. There are three reasons why the blockchain technology can secure
the transmission as follows:
1. It has an immutable ledger that people are allowed to manage and
control. Once the record is saved in the ledger, it is not changeable. Moreover,
the ledger should follow a prebuilt regulation.
2. The blockchain technology has been distributed and is able to operate
at the same time from various devices and computers.
3. The blockchain also follows the regulation and data exchange policies
including a smart contract system. The smart contract system handles identity
and access permission of electronic medical reports (EMRs) which are saved in
30
the blockchain. It only gives out permission to doctors to go through EMRs for
their patients. The healthcare data gateway (HDG) through an app utilizes
blockchain technology (Yue et al., 2016). It provides authorities to patients for
saving their medical information along with the privacy policy.
C. IoT Healthcare Applications
IoT healthcare applications are used and focused by the patients.
Therefore, healthcare applications are more toward the user based. In this
section, IoT healthcare applications are covered such as Electrocardiogram
(ECG) Monitoring, Glucose Level Sensing, Blood Pressure Monitoring, Body
Temperature Monitoring, Oxygen Saturation Monitoring, Asthma Monitoring,
Mood Monitoring, Medication Management, Rehabilitation System, and
Wheelchair Management.
Electrocardiogram Monitoring (ECG). Electrocardiogram (ECG) describes
the heart's electrical activity by reason of depolarization and repolarization of
atrium and ventricles (Pradhan et al., 2021). ECG provides fundamental rhythms
of the heart muscles. It also indicates abnormalities of cardiac symptoms such as
arrhythmia, myocardial ischemia, and prolonged QT interval (Drew et al., 2004).
The utilization of IoT technology has discovered the capacity of application in
detecting early heart abnormalities with ECG monitoring application. One study
has found out an IoT-based ECG monitoring application which is using receiving
processors and wireless data acquisition systems (Liu, 2012). It used a search
31
automation method for detecting cardiac abnormalities on a real time basis. A
low-powered small ECG monitoring system that was integrated with a t-shirt had
a biopotential chip for collecting ECG data (Wu et al., 2019). The earned data is
transferred to the end-users with a Bluetooth function, and it can be visualized
with a mobile app. An ECG monitoring system can operate long-term and
continuous monitoring through the incorporation of nanoelectronics, IoT, and big
data (Bansal & Gandhi, 2019). A compressive sensing optimizes the power
consumption and conducts a better ECG monitoring performance (Djelouat et al.,
2020). ECG monitoring system and fall detection using IoT technology can keep
track of elderly patients’ health conditions (Al-Kababji et al., 2019). It uses a
cloud-based server and integrates with a mobile application.
Glucose Level Sensing. Glucose Level is the most important measure for
people who are suffering from diabetes. Diabetes is a metabolic disease that
causes high blood glucose levels in a prolonged period. Monitoring glucose level
shows each pattern of blood glucose level so that patients can prepare for their
meals, medication, and activities according to their blood glucose level. As IoT
technologies grow, there are different kinds of wearable gadgets that have been
developed for monitoring blood glucose level using IoT technologies. IoT-based
noninvasive real-time glucometers for monitoring blood glucose levels in
wearable sensors are linked to the healthcare providers through IPv6
connectivity (Istepanian et al., 2011). A glove for evaluating blood glucose level
uses a Raspberry Pi camera with a laser beam (Alarcón-Paredes et al., 2019).
32
With a Raspberry Pi camera, there are pictures of fingertips for analyzing the
diabetic condition of the users. An algorithm measures glucose level by
employing double moving average of IoT architecture (Valenzuela et al., 2020).
In addition, optical sensors use light signals reflected from the patient's body to
calculate the glucose level from the human body.
Blood Pressure Monitoring. Blood Pressure is a mandatory test for most of
the diagnostic procedures at the hospital. Wearable cuffless gadgets can
measure systolic and diastolic pressure (Xin & Wu, 2017). The data monitored by
a wearable cuffless gadget is saved in the cloud system. Moreover, Fog
computing and cloud computing has been used for measuring Blood Pressure in
IoT-based systems (Guntha, 2019). Using fog computing and cloud computing
make systems to monitor Blood Pressure in long-term and real-time manner.
Also, Convolutional Neural Networks (CNN) models based on a deep learning
system can evaluate the systolic pressure and the diastolic blood pressure (Chao
& Tu, 2017). Blood pressure measurement system uses Photoplethysmography
(PPG) and the ECG signal (Dinh et al., 2017). The blood pressure was analyzed
by the attached microcontroller module. It also stored the recorded data in the
cloud storage.
Body Temperature Monitoring. Body temperature is an important part of
any diagnostic procedure in healthcare services since body temperature provides
a vital sign of maintenance of homeostasis (Ruiz et al., 2009). In addition, a little
33
change in body temperature can affect the human body and can be fatal.
Therefore, it is important to keep track of body temperature for doctors to decide
about a patient's health condition for further treatment. Using a temperature
thermometer is a typical way of measurement. However, IoT-based technologies
have developed a way to monitor body temperature in a more comfortable and
stable way. 3D-printed wearable devices can be put on the ear (Ota et al., 2017).
The tympanic membrane in the ears makes a device to check the core body
temperature using an infrared sensor. This device is combined with a wireless
sensor and data processing system. Moreover, the device is not affected by the
surrounding environment or by any other physical activities so that it can
measure the human body temperature with integrity. IoT-based temperature
monitoring system that can store the data in the database and be able to display
on a web page allowing people to access it with their desktop or a mobile phone
(Gunawan et al., 2020). The system has been developed using Arduino and
Raspberry Pi. In addition, wearable devices with a lightweight sensor for
measurement of an infant's body temperature in real-time could be reported to
their parents when a critical change of temperature happens (Zakaria et al.,
2018).
Oxygen Saturation Monitoring. Pulse oximetry is a proper way of oxygen
saturation measurement and a crucial parameter in healthcare analysis. It is the
noninvasive method which reduces conventional issues from previous methods
and makes it possible to do real-time monitoring. Integrating IoT-based
34
technology and pulse oximetry has improved the capability of healthcare
application. An alarm system can notify a patient's oxygen saturation level when
it reaches a critical point (Agustine et al., 2018). The alarm system was combined
with a WLAN router using the Blynk server and pulse oximeter. Also, noninvasive
tissue oximeters can calculate the blood oxygen saturation including pulse
parameters and heat rate (Fu & Liu, 2015). In addition, monitored data can be
transmitted through the server using communication technologies so that the
data is used for doctor’s reference about the patients.
Asthma Monitoring. Asthma is one of the hardest chronic diseases that
can cause fatal effects on a person's breathing and airways. When airways
shrink, it causes critical health problems such as shortness of breath, chest pain,
wheezing, and coughing. The only way of controlling asthma is using an inhaler
or nebulizer when it happens. IoT-based smart sensors monitor respiratory rate
in asthma patients (Shah et al., 2018). The monitored data of patients also was
saved in a cloud server that allows caregivers to access diagnostic and
monitoring objectives. A respiratory monitoring system with an alarm system
uses a LM35 temperature sensor for measurement of the respiratory rate (Raji et
al., 2016). This system was conducted by observing the air of inhalation and
exhalation from the patients. The monitored data were transmitted to the health
center and can be seen on a web server. In addition, this system automatically
transmitted an alarm to the patient when a threshold level was reached. A
system is developed which is able to monitor their condition, warn the health
35
condition of the patients, and suggest a proper medication to be administered
(Gundu, 2020). Moreover, the system analyzed the environmental condition for
the patient's health and directed the patients to change their position for their
health. IoT-based devices integrated with machine learning, big data analysis,
and cloud computing has suggested to monitor asthma more effectively (Prasad,
2020). Most potential features of monitoring asthma in the near future are
utilizing the IoT-based asthma monitoring system (Hui et al., 2021).
Mood Monitoring. Mood monitoring is tracking of a person’s emotional
state that can be used for stabilizing a healthy mental state. Specifically, mood
monitoring can be used for dealing with mental diseases, and it also supports
healthcare professionals to manage their patients who are having a bipolar
disorder, depression, stress, and so on. In addition, it is helpful for people to
understand their mental condition by self-monitoring their emotional state. Mood
mining approach using a CNN network for evaluating and categorizing individual
moods in 6 categories which are happy, sad, calm, excited, distressed, and
angry (Alam et al., 2017). Meezaj is an interactive system for measurement of
real-time mood (Ahmad, 2020). Also, there is a system that can communicate
with the people regarding their stress level (Pandey, 2017). The stress level can
be utilized in designing IoT-based systems so that people prevent any further
acute condition from occurring. MoodRecord is a platform that can monitor
patients with bipolar disorder using Smartphones at home (Codina-Filbà et al.,
2021). It is recorded user activity and requires users to answer questions and
36
asks users to record video of themselves that is designed by physicians (Codina-
Filbà et al., 2021).
Medication Management. Medication noncompliance causes a threat to
personal health and public health. It also generates huge financial waste all over
the world. As people are getting older, medication nonadherence occurs along
with individual conditions such as dementia and cognitive decline. Many research
has been developed for tracking the patient’s medication compliance by using
IoT applications. A smart medical box reminds people to take their medicine
(Bharadwaj et al., 2017). In addition, the smart medical box can calculate the vital
health parameters such as temperature, ECG, blood oxygen level, and blood
glucose level. The recorded data is transmitted to the cloud server and can be
accessed by patients and doctors through the mobile app. There is a system that
monitors the condition of medication storage such as humidity and temperature
so that the medication can be maintained in an ultimate storage environment for
medicine (Karagiannis & Nikita, 2020). Saathi is a medication monitoring system
for women who are in vitro fertilization (IVF) treatment (Wadibhasme et al.,
2020). It is especially critical to take medication on time for women who are in the
IVF process.
Rehabilitation System. Rehabilitation system is an essential physical
medicine for restoring the functional ability of the patients who are suffering from
disability (Pradhan et al., 2021). The Rehabilitation system is designed to identify
37
the problem and assist patients to return to their normal life. The application of
IoT with rehabilitation can be used for the treatment of cancer, stroke, sports
injury, and physical disabilities (Qi et al., 2018). A smart walker rehabilitation
system monitors the patient’s walking pattern using a multimodal sensor and
then analyzes the movement metrics (Nave & Postolache, 2018). The system
was also utilized with a mobile app so that doctors have access to the recorded
data and provide diagnostic reports. In addition, a sports rehabilitation system
that can monitor the temperature, electromyography (EMG), motion posture, and
electrocardiogram (ECG) has been developed and gives feedback to the athletes
(Pradhan et al., 2021). The monitored data can be used in healthcare
professionals for prediction of a patient's recovery process and plan of further
rehabilitation program (Pradhan et al., 2021).
Wheelchair Management. A wheelchair assists patients with restricted
mobility in a physical and psychological way. However, using a wheelchair can
be limited to the patients who have brain damage. IoT-based steering system
utilizes an obstacle avoidance system in a real-time basis (Lee et al., 2017). The
system was designed to use image processing techniques of the recorded real-
time basis videos. A smart wheelchair was utilized by multiple sensors, cloud
computing, and mobile technologies (Ghorbel et al., 2018). A smart wheelchair
system has a mobile app for active interaction between patients and the
wheelchair, the caregivers. Moreover, it allows caregivers to monitor the
wheelchair movement from a distance. IoT-based wheelchair monitoring system
38
makes it possible to control with hand gestures (Garg et al., 2018). It is a useful
controlling system for those who are having quadriplegia. The system used RF
sensors in hand gloves for controlling the wheelchair. In addition, the information
collected by the sensors was transmitted to the server for storing in the cloud
which enables doctors and caregivers to access the data and refer to the
patient's diagnosis. Advanced version of smart wheelchair can monitor the
wheelchair activity and provide obstacle detection features, head mat, foot mat,
and an umbrella (Kumar et al., 2020).
D. IoT Healthcare Services Focusing on Smartwatch
About 20% of Americans currently own smart wearable devices, thus the
global market is expected to reach $70 billion by 2025 with a 25% of annual
growth rate (Polaris Market Research, 2020). Among the variety of smart
wearable devices, Smartwatches have been developed with the combination of
mobile technology and medical devices so that Smartwatches can monitor users’
health including cardiovascular parameters. There are numerous Smartwatches
that have been developed. According to a report conducted by Counterpoint
Research, the following numbers are Smartwatches global market share in 2021:
Apple occupies 33.5%, Huawei 8.4%, Samsung 8%, iMoo 5.1%, Fitbit 4.2%,
Other 40.8% (Lim, 2021). In this section, Apple watch and Samsung Galaxy
watch will be compared.
39
Apple Watch. According to the Apple website
1
The Apple Watch 7
provides heart rate notifications, irregular rhythm notifications, fall detection, and
has an electrocardiogram (ECG) app which has a heart sensor (Apple, 2021).
The ECG has been approved by the Food and Drug Administration (FDA) for
determining atrial fibrillation (AF) symptoms (Walters et al., 2016). The ECG app
can monitor atrial fibrillation and sinus rhythm and it also asks users to input
health symptoms. The monitored data can be shared with healthcare providers.
Hence, if the user has serious symptoms, the ECG app prompts them to call
emergency services. The ECG app can only be used to obtain information and
medical consultation through the devices with personal healthcare providers
before taking any action. Therefore, users can keep monitoring their heart
condition before going to see a doctor and after. The Apple watch also uses
Photoplethysmography (PPG) which can do continuous monitoring and has an
algorithm that allows users to detect atrial fibrillation (AF) symptoms by
themselves (Isakadze & Martin, 2020). PPG is generated by a pulse oximeter
which lights the skin and measures any change in light absorption and is useful
for measuring the heart rate. Through the PPG technology can monitor a
tachogram showing heartbeats and apply it to the algorithm so that it can detect
the pulse irregularity and AF. However, using PPG to monitor AF is not approved
by the FDA (Dickson et al., 2021).
1
http://www.apple.com/healthcare/apple-watch/
40
Samsung Galaxy Watch. According to the Samsung website
2
The
Samsung Galaxy Watch 4 can monitor users’ blood oxygen level (SpO
2
), heart
rate, steps, distance, calories, sleep stages (Awake-REM-Light-Deep), active
time, and stress (Samsung, 2021). The Samsung Galaxy Watch can also
measure users’ weight, skeletal muscle, fat mass, body fat, BMI, body water, and
BMR (Westenberg, 2021). Samsung employs BioActive Sensor (including PPG,
ECG, and bioelectrical impedance (BIA) sensor) which enables the Galaxy
Watch to calculate the ECG in a real time basis so that users can monitor their
abnormal heart rate using the ECG and share the monitored result with their
phone. It also manages users' sleep quality since it tracks users’ sleep stages,
including snoring and blood oxygen levels while users are sleeping. The
Samsung Galaxy Watch also can be used in tracking users’ working out routine
and burning calories with GPS. Table 4 shows the comparison between Apple
Watch and Samsung Galaxy Watch (Apple, 2021; Westenberg, 2021).
Table 4. Comparison between Apple Watch and Samsung Galaxy Watch
Feature
Apple
Samsung
Model
Apple Watch Series 7
Galaxy Watch 4
Release
Month
October 2021
August 2021
Height
41mm, 45mm
40mm, 44mm
Width
35mm, 38mm
39.3mm, 43.3mm
Depth
10.7mm
9.8mm
2
https://www.samsung.com/us/watches/galaxy-watch4/
41
Weight
32g-51.5g (depending on
material)
25.9g, 30.3g
Display
Retina LTPO OLED
AMOLED
Display size
1.4-inch 1.77-inch
1.19-inch, 1.36-inch
Features
GPS, GLONASS, Galileo,
QZSS, and BeiDou
Compass
Always-on altimeter
Water resistant
50 meters
Blood oxygen sensor
(Blood Oxygen app)
Electrical heart sensor
(ECG app)
Third-generation optical heart
sensor
International emergency calling
Emergency SOS
Accelerometer
(Up to 32 g-forces with fall
detection)
Gyroscope
Ambient light sensor
Speaker
Microphone
Apple Pay
GymKit
GPS, GLONASS, Beidou, and
Galileo
Accelerometer
Barometer
Gyroscope
Geomagnetic sensor
Ambient light sensor
Samsung BioActive sensor:
optical heart rate (PPG),
electrocardiogram (ECG),
bioelectrical impedance
analysis sensor (BIA)
Ambient Light
HRM
Samsung Pay
Ability
Blood Oxygen, Heartrate, ECG,
Irregular rhythm notifications,
Fall detection, Steps,
Distances, Speed, Calories,
Exercise, Sleep, Elevation,
Active time, Respiratory rate
Blood oxygen level (SpO2),
Heart rate, Steps, Distance,
Calories, Sleep stages
(Awake-REM-Light-Deep),
Active time, Stress, Weight,
Skeletal muscle, Fat mass,
Body fat, BMI, Body water,
and BMR
(Apple, 2021; Westenberg, 2021)
42
As we discussed IoT healthcare services, service enablers and
applications in this chapter, IoT has the ability to overcome human limitations and
physical difficulties. Among the variety of IoT services, Smart Watches have
been considered as commonly used wearable technology with combination of
Smartphones to help peoples’ health monitoring. As we can see the usefulness
of health monitoring through wearable devices, it implies that the integration of
IoT and Health Recommender Systems would create more possibilities in human
life.
43
CHAPTER FOUR:
HEALTH RECOMMENDER SYSTEMS
Integration of IoT and Recommender Systems in healthcare can support
people to pay more attention to their personal health and wellness.
Recommender Systems are aimed to assist people to make decisions when they
are short on knowledge where there are several options available to choose from
(Ricci et al., 2011). Recommender systems are utilized in predicting the user
preferences of certain items (Bobadilla et al., 2013). The recommender systems
have been used for helping a process of decision making in e-commerce,
transportation, healthcare, agriculture, and media (Fayyaz et al., 2020). In
particular, the Health Recommender Systems provide medical information that is
related to the patient’s health improvement and medical treatment along with
patients’ health records. Health Recommender Systems contribute to the
advancement of healthcare services by offering patient centric health information
to the healthcare professionals and patients in a proper moment. Moreover,
Health Recommender Systems deliver the patient’s medical information which is
monitored data from wearable devices and personal medication history to the
Clinical Diagnosis System (CDS) so that the patient can have a recommendation,
such as diagnosis, medication refills, health insurance plans, follow-up, drug
interaction alert, and diet recommendations regarding their health. Health
Recommender Systems in IoT improve convenience feature of Health
44
Recommender System to the users. For example, the Holter monitor is one of
the portable electrocardiograms (ECG) devices that can do remote monitoring
regarding heart related disorders (Sana et al., 2020). However, it is difficult to
carry around due to the big size, thus the usage has been reduced. Regarding
the size problem, wearable devices have been represented for its convenience
and remote monitoring in a long-term period. One of the advantages of IoT-
based Health Recommender Systems can be the convenience for patients to
monitor their health with less effort.
The core concept of a recommender system can be shown using the
following function below (Adomavicius & Tuzhilin, 2005):
f: R × I → X
R
means all the users and
I
means possible recommended items. Where f
implies that utility of function about a certain item
i∈I
to a user
r∈R. X
refers to the
final recommendation including some of items that users have not used before
but might like. Recommended items were listed by expanding the utility function.
In addition, the following formulation can be used (Adomavicius & Tuzhilin,
2005).
∀r ∈ R, i’ᵣ= arg max
f (r, i).
i∈I
Above the formula implies that to all the users
r∈R
, the recommender
system decides to choose a certain item
i’ᵣ∈I
which can maximize the utility of
45
function to the users. The result of predicting the utility of certain items is
depending on the selected recommendation algorithms. There are four main
recommendation techniques: content-based filtering, collaborative-based filtering,
knowledge-based filtering, and hybrid recommendation systems which are
discussed below (Zhang et al., 2020).
Main Recommender Systems and Algorithms
Content-Based Filtering
Content-based filtering employs an item’s description for forecasting its
utility along with user’s preferences (Shardanand & Maes, 1995). Content-based
recommender systems pursue a goal that recommends similar items based on
an individual user’s previous interest. In formal terms, the utility
f (r, i)
of certain
items
i
for a user
r
is calculated by the utility
f (r, iᵥ)
from a user
r
to certain items
iᵥ∈I
that is similar to item
i
.
Most commonly used retrieval technique is using a
keyword-based model, called the vector space model (Salton et al., 1975).
Content-based recommender systems follow user’s consumption records to find
a user’s interest on specific items and store them in the user's profile. The
profiling process can encounter a binary classification problem. At this step,
classic methods can be used such as Naïve Bayes, decision trees and nearest
neighbor algorithms (Sebastiani, 2002). Content-based recommender system
filters and matches the item representation with the user profile (Zhang et al.,
2020). Moreover, the result is forwarded and the unmatched items that users
46
dislike is removed. Content-based recommender system is on the basis of item
representation and individual user. Therefore, it has enough information to
predict the user's preferences. Also, this system has no cold-start problem since
it can recommend new items to the individual users. In addition, this system is
able to explain clearly about the recommendation result. However, the content-
based recommender system is having a problem which are new user issues,
over-specialization, data sparsity, and scalability (Balabanović & Shoham, 1997).
Besides, it encounters the limitation to the variety of recommended items which
means it keeps recommending the similar item for users. Furthermore, the items
are not able to be represented in a specific form which is required to use a
content-based recommender system.
Collaborative-Based Filtering
Collaborative-based filtering is aimed to predict unknown results by
making a user-item rating matrix depending on a user’s item preferences or
choices (Sahoo et al., 2019). The users who have not ever rated certain items
will get the recommended items according to the positive rating by other users. In
formal terms, the utility
f (r, i)
of a certain item
i
is calculated by the similar users
r'∈R
experienced the utility of the assigned item
f (r', i)
. Collaborative-based
filtering systems can be divided into memory-based collaborative-based filtering
and model-based collaborative-based filtering (Zhang et al., 2020). Memory-
based collaborative-based filtering uses the nearest neighbor algorithm. The
47
recommendation computes possible ratings of certain items depending on the
user’s neighborhood user or item. In addition, memory-based collaborative-based
filtering uses heuristic algorithms for calculating similarity values on users or
items and it is also divided into user-based collaborative-based filtering and item-
based collaborative-based filtering (Deshpande & Karypis, 2004). The memory-
based collaborative filtering is easy to implement and an effective application.
However, it has a cold-start problem. So, it is hard to predict if the user and items
are new on the system. Even if an item is not a new product but unpopular, it is
hard to get a rating from the consumers. It is especially hard to offer a
recommendation in a real-time basis since the heuristic process needs time to
provide a recommendation. This drawback is partially figured out by item-based
collaborative filtering using a pre-calculated and pre-stored matrix. Model-based
collaborative-filtering utilizes a machine learning/ data mining technique. When
the ancillary information is integrated with the rating matrix, the model-based
collaborative-filtering system brings out great outcomes. Matrix factorization won
the Netflix Prize in 2009 and it has been ranked as the most popular algorithm
(Koren et al., 2009). Matrix factorization can solve the sparsity problem (Luo et
al., 2016). Even though users rated only a few items, they still can acquire an
accurate recommendation using the matrix factorization. In addition, matrix
factorization can integrate with other information so that it is easy to finish up the
user profile and assist recommender systems performance (Liu et al., 2015).
48
Due to the shortcomings of content-based filtering such as cold-start
problems, scalability, over-specialization, and data sparsity, there are studies to
overcome the problem combining with Naive Bayes classification (Ghazanfar &
Bennett, 2010). Therefore, it increases the accuracy and coverage of
experimental results by integrating the collaborative filtering and Naive Bayes
classification. In addition, the Bayesian model with collaborative filtering resulted
in providing a recommendation which shows good prediction as matrix
factorization (Valdiviezo-Diaz et al., 2019). Collaborative filtering with decision
trees is also being used with dimensionality reduction and separate decision
trees in every attribute (Wu, 2019). Moreover, using Reversed collaborative
filtering (RCF) and K-Nearest Neighbors (KNN) perform accurate predictions with
filtering inaccurate results and help to quickly find similar users. RCF is based on
finding rated items using the KNN graph (Park et al., 2015). Collaborative filtering
with KNN and gradient boosting method is developed (Lu et al., 2018).
Collaborative filtering with KNN enhances the calculation of similarity of items
and users for obtaining more relevant information and Collaborative filtering with
gradient boosting method is also used for predicting the items’ score.
Knowledge-Based Filtering
Knowledge-based filtering is on the basis of existing knowledge, regulation
about item functions and user needs (Burke, 2002). Knowledge-based
recommender system possess knowledge which is extracted from previous
records of the user. The knowledge includes constraints, problems, and solutions
49
(Zhang et al., 2020). Case-based reasoning employs previous cases to resolve
the current problem using the knowledge-based system (Aamodt & Enric, 1994).
With knowledge-based filtering system, it finds similarities among the products,
and it requires organized representations. The knowledge-based recommender
system can especially be used in health decision support, financial services, and
house sales since these services need certain domain knowledge and specific
situations (Felfernig et al., 2011). Moreover, a knowledge-based recommender
system does not suffer from having a new item/user problem because this
system is storing the knowledge already (Felfernig & Burke, 2008). Moreover,
users can make constraints about the outcome of recommendation. The
disadvantage of this system is that the knowledge-based recommender system
costs high to set up and manage the knowledge base system (Zhang et al.,
2020).
Hybrid Filtering
Hybrid filtering system comprises collaborative-based filtering and content-
based filtering for enhancing the performance and accuracy of the recommender
system (Sahoo et al., 2019). Content-based filtering systems do not include all
the opinions for recommending items (Chavan et al., 2021). In addition, it has a
limitation of providing a recommendation if it is in the user’s interest.
Collaborative-based filtering has a weakness that the systems are not able to
provide recommendations if the item has not been rated which is called a cold-
50
start problem. On the other hand, hybrid filtering systems can overcome the
limitations from content-based filtering and collaborative-based filtering.
Moreover, it consists of a combination of different recommender techniques so
that it will increase the accuracy and optimal recommendations compared to a
single recommender system (Schafer et al., 2007). In the healthcare perspective,
a user profile can be extracted from a Personal Health Record (PHR). With using
a PHR, it can solve the cold-start problem in the collaborative filtering system.
Hybrid filtering is aimed to earn more accurate results and enhance the
performance of algorithms. Hybrid filtering has been divided into seven
strategies, such as weighted, mixed, switching, feature combination, feature
augmentation, cascade, and meta-level filtering (Fayyaz et al., 2020).
Health Recommender Systems have been developed and immersed into
our daily life to enhance our health and life through the utilization of filtering
technologies. For example, sending motivational messages about the possibility
of making a difference in their behavior using hybrid filtering recommender
systems is helpful for smokers to quit smoking (Hors-Fraile et al., 2016). In
addition, Health Recommender Systems have been developed to make
recommendations for dietary needs depending on the patients’ medical history
and body features using machine learning algorithms and deep learning
algorithms such as Long Short-Term Memory (LSTM), recurrent neural network,
and Naive Bayes (Iwendi et al., 2020). For psychological recommender systems,
51
it has been suggested that knowledge-based recommender system recommend
patients for naturopathy, music therapy, and art therapy so that they can improve
their psychological symptoms (Gyrard & Sheth, 2020). Moreover, it was possible
to find rare diseases using a hybrid recommender system which is a combination
of collaborative-based filtering and context-based filtering (Almeida et al.,
2020).
Machine Learning Algorithms
There are different machine learning algorithms used in Recommender
System to predict and filter for identifying the disease depending on the data type
and needed implementation results. Support Vector Machine (SVM) is one of the
supervised learning models. This model has been used for analyzing data for
regression and classification analysis. In particular, SVM efficiently implements
non-linear and linear data by using the kernel types which are mapping the inputs
to the high-dimensional spaces (Mahesh, 2020). In addition, SVM-based
collaborative filtering has been proposed to improve the precision and efficiency
of recommendations (Chang et al., 2019). There is multi-criteria collaborative
filtering with SVM which enhances precision of recommendation (Nilashi et al.,
2014).
The Decision Tree (DT) algorithm is also one of the supervised machine
learning algorithms (Ibrahim & Abdulazeez, 2021). DT keeps dividing the dataset
until it solves classification or regression problems. The tree is made by a training
52
process and final leaf nodes means the final decisions. The DT algorithm is also
used to detect breast cancer.
The Naive Bayes (NB) algorithm is a classification algorithm that uses a
statistical method and probability method (Yaswanth & Riyazuddin, 2020). NB
algorithm is utilized in the classification of documents and filtering emails. This
machine learning algorithm is characterized by its simplicity and usefulness with
large amounts of data in most of the field.
The K Nearest Neighbors (KNN) algorithm uses Euclidean distances
between data to earn neighbors (Urgiriye & Bhartiya, 2020). It is one of the most
widely used algorithms to resolve any regression and classification issues. The K
value can discover the similar cases that exist for the new case and find a similar
category with the new case. Therefore, the value of K should be decided
carefully to not result in overfitting.
The Random Forest algorithm is an ensemble model used to predict the
nearest neighbors and the main idea for the ensemble is that a group of models
forming a strong model (VijiyaKumar et al., 2019). The benefit of random forest
classifiers is that they have concise running time and can handle missing data
and unbalanced data. Subtree created by the dataset can choose the class in the
dataset. Random forest algorithms can be used for diagnosing heart disease
(Meshref, 2019).
The Logistic Regression algorithm is one of the supervised learning
algorithms which can resolve in the formatting of binary classification by
53
employing mathematical models with logistic function which is a sigmoid function
for modeling the data (Kumar, 2020). Logistic Regression is characterized by
simplicity of execution, training-based effectiveness, ease of regularization, and
computational efficiency.
The Extreme Gradient Boosting (XGBoost) algorithm is applied for
classification and prediction of problems due to the efficiency, portability, and
flexibility of the algorithms (Kao et al., 2020). XGBoost is based on gradient
boosting. XGBoost algorithms solve the over-fitting problem and improve the
computational resources. XGBoost can draw the result by simplification of the
objective functions with the high computational speed (Fan et al., 2018).
In this chapter, Recommender Systems are elaborated with highlighting
Healthcare Recommender Systems. There are four main Recommender
Systems which are content-based filtering, collaborative-based filtering,
knowledge-based filtering, and hybrid filtering. The usage and applications of
Health Recommender Systems does not have an adequate number of studies in
the past. However, Health Recommender Systems have the potential to expand
the scope of improving human life. Moreover, various machine learning
algorithms which can be used in recommender systems have been reviewed in
this chapter. Therefore, an IoT-based Health Recommender Systems is
suggested in the next chapter based on the investigated machine learning
54
algorithms. The suggested framework also includes utilizing the Middleware-
based IoT architecture for the purpose of integration.
55
CHAPTER FIVE:
CASE STUDY AND DISCUSSION
The Internet of Things (IoT) and Health Recommender Systems are
presented in the previous chapters. The importance and advantage of integrating
IoT and Health Recommender Systems are also presented which can enhance
the quality of human life and wellness. IoT can broaden the scope of Health
Recommender Systems usability for various healthcare areas. In addition, a case
study is presented in this chapter to indicate how different machine learning
methods can be used for Health Recommender Systems and how effective they
are for the case under study.
IoT-Based Health Recommender Systems Framework
In this chapter, an IoT-based Health Recommender Systems Framework
is suggested. The framework uses the Middleware-based IoT architecture. Figure
3 shows the proposed IoT-based Health Recommender Systems Framework
developed in this study.
56
Figure 3.
IoT-based Health Recommender System
s framework
57
The framework is based on Middleware-based IoT architectures, so it has
5 layers: perception layer, network layer, middleware layer, application layer, and
business layer. As Figure 3 shows, in the perception layer the system will start
with sensors health data from IoT devices and wearable devices such as
Smartphone, Smartwatch to perceive important medical parameters, for
example, heart rate, heart variability, blood pressure, blood sugar, oxygen
saturation, respiration rate, pulse rate, body temperature, and any symptoms
input from patients. This information from users is being monitored and saved in
the cloud server using the network layer. In the network layer, MQTT and CoAP
can be used depending on needs. Message Queue Telemetry Transport (MQTT)
is a lightweight messaging protocol that can deliver messages from remote
locations (Ansari et al., 2018). Constrained Application Protocol (CoAP) is an
internet application protocol for constrained devices, and it allows constrained
devices to communicate with the wider internet which uses similar protocols. In
the middleware layer, monitored data, drug reviews, and medical history for each
individual will be saved in the cloud server and repository to check whenever it is
needed. Moreover, the cloud server and repository can be shared with users and
healthcare providers since it is important to know about critical information for
both. In the application layer, the most important analytic software is being
executed. All the data recorded from users and information in cloud servers are
being analyzed in this layer using Recommender System, Machine Learning
algorithms, and big data analytics. In addition, the results from the analytic
58
engine and recommendation system should be reviewed and advised by the
doctors and health professionals to proceed to the next layer. The business layer
can be divided into two categories which are critical and non-critical situations
depending on the results from the application layer. If the user’s health condition
is critical, IoT devices will recommend to the emergent patient to see a doctor,
report to their patient’s family or caregivers, and send over an ambulance to
patients or deliver a medicine from the hospital through the quick delivery
medium. Otherwise, it would notify the user lifestyle recommendations, nutrition
recommendations, and general health information to the non-critical situation
patient through the IoT devices.
Case Study
Various machine learning algorithms have been reviewed in the previous
chapter. The types of algorithms which show higher accuracy performance are
implemented using Python in this chapter. In addition, the most important
features for an occurrence of heart diseases are discussed by using a data
source from the University of California, Irvine (Janosi et al., 1988). The dataset
consists of 303 observations and 14 variables. 14 variables are age, sex, chest
pain type, resting blood pressure, serum cholesterol, fasting blood sugar, resting
electrocardiographic results, maximum heart rate, exercise induced angina, old
peak, slope of the peak exercise ST segment, number of major vessels,
thalassemia value, target. 303 observations vary the age range between 29 to
59
77. The accuracy output on each model is Support Vector Machine and K-
Nearest Neighbors: 88.52 %, Random Forest: 86.88%, Logistic Regression,
Extreme Gradient Boosting, and Naive Bayesian: 85.24%, and Decision Tree:
81.96%. In this case study, seven machine learning methods including Support
Vector Machine, Decision Tree, K-Nearest Neighbors, Random Forest, Extreme
Gradient Boosting, Naive Bayesian, and Logistic Regression are implemented to
develop different Health Recommender Systems for the patients with heart
diseases. Then, the accuracy of each recommendation system has been
calculated to show the effectiveness of each method for the under-study dataset.
Appendix A includes the Python code of all seven implemented machine learning
methods. Figure 4 shows accuracy of different machine learning algorithms that
are being used in the case study based on the heart diseases dataset. The
accuracy is calculated by comparing the model output on the test data with the
pre-known labels of data (i.e., 0 means the patient does not have heart diseases,
1 means the patient does have heart diseases). For each algorithm the accuracy
is calculated to show how many times the model can predict the output correctly.
The values are multiplied by 100 to be shown in percentage.
60
Figure 4. Accuracy of different Machine Learning algorithms
According to the barplot in Figure 4, Support Vector Machine (SVM)
algorithms and K Nearest Neighbors (KNN) are both the same with the highest
accuracy, 88.52%. This means that both algorithms can predict the possibility of
heart diseases from the patients. SVM has less of a possibility of overfitting and
can manage linear data and non-linear data. SVM has weaknesses in dealing
with large quantities of dataset. The implemented data set subject was 303 and
14 variables, so SVM can achieve high accuracy in this case study result. On the
other hand, KNN has strength in multiclass problems and is used for
61
classification and regression problems. Even though KNN has high sensitivity to
irrelevant and high computation, it showed high accuracy in this study.
Figure 5. Feature importance for heart diseases
Figure 5 shows what features have an impact on the occurrence of heart
diseases. The feature importance is calculated by using feature importance
(except the target data) using a XGBoost. The conducted features include the
following: age, sex, chest pain type (cp), resting blood pressure (trestbps), serum
cholesterol (chol), fasting blood sugar (fbs), resting electrocardiographic results
62
(restecg), maximum heart rate (thalach), exercise induced angina (exang), old
peak, slope of the peak exercise ST segment (slope), number of major vessels
(ca), thalassemia value (thal). Among these features, thalassemia value,
exercise induced angina, and chest pain type are the most influential
components to heart diseases. ECG should be used in most heart disease cases
to proceed with treatment or testing. Therefore, ECG technology in wearable
devices allow users to monitor the irregular rhythm of their heart.
Figure 6. Correlation between the features of heart diseases
63
Figure 6 represents the correlation between the features of heart
diseases. The most correlated value is 0.39 which shows correlation between
maximum heart rate (thalach) and the slope of the peak exercise ST segment
(slope). ST segment is elevated by early repolarization, acute ischemia,
ventricular dyskinesia, and injury from pericarditis (Kashou et al., 2021).
In this chapter, an IoT-based Recommender System framework has been
suggested and the case study was implemented showing the accuracy of
machine learning algorithms to predict heart diseases and the feature importance
of heart diseases and correlation between the features of heart diseases. SVM
algorithms and KNN algorithms resulted in higher accuracy compared to other
algorithms to predict heart disease occurrences rates. Such methods can be
utilized in the proposed IoT-based framework to recommend patients appropriate
actions based on their situation. For example, if the recommender system
classifies the patients in high risk of heart diseases it will call the emergency and
notify the family of the patient. In case of non-emergency, the recommender
system will specify what the patient should do to avoid any future discomfort. The
system is capable of finding important features for patients with possible heart
diseases (such as thalassemia value, exercise induced angina, and chest pain
type) and recommend actions to improve their health and be conscious of their
condition. The proposed framework helps address some challenges, for
example, the risk of going to the doctor in case of pandemic, taking quick actions
64
in case of emergency conditions, affordability of healthcare services, and
improving the living habits by considering recommendations in non-emergency
situations.
65
CHAPTER SIX:
CONCLUSION AND FUTURE WORK
Conclusion
This project is aimed to develop an IoT-based Recommender System
framework for coupling IoT and Health Recommender Systems. This project
presents the architecture of smart IoT in healthcare, IoT healthcare services, IoT
healthcare service enablers, IoT healthcare applications, and IoT healthcare
devices focusing on wearable devices, especially a Smartwatch. In addition, this
project describes the Recommender Systems with machine learning algorithms
which is applicable to the Health Recommender Systems in many ways. The
integration of IoT and Health Recommender Systems will enhance people's
wellness, detect diseases earlier and provide medical advice to patient’s IoT
devices or wearable devices. This case study is conducted with data about heart
diseases. Through the case study, the accuracy of machine learning algorithms
regarding health-related issues are high in Support Vector Machine (SVM) and
K-Nearest Neighbors (KNN) algorithms. This study can be helpful for people who
need to develop a health-related recommender system using machine learning
algorithms. The feature importance for heart diseases and the correlation
between the features of heart diseases can support people who need information
regarding long term actions to prevent heart diseases.
66
As the project reaches the finish line, answers to each research question
are provided in this section. This project investigated the IoT healthcare services
which are Ambient Assisted Living, mHealth, Wearable Devices, Adverse Drug
Reaction, Community Healthcare Monitoring, and Children Health Information.
IoT healthcare service enablers are Cognitive Computing and Blockchain. In
addition, IoT healthcare applications are ECG, Glucose Level Sensing, Blood
Pressure Monitoring, Body Temperature Monitoring, Oxygen Saturation
Monitoring, Asthma Monitoring, Mood Monitoring, Medication Management,
Rehabilitation System, and Wheelchair Management for enhancing people’s
wellness and overcoming physical limitations. Content-based filtering,
collaborative-based filtering, knowledge-based filtering, and hybrid filtering are
described with machine learning algorithms for the understanding how the Health
Recommender Systems work. Also, this study shows the usage of Health
Recommender Systems in helping to quit smoking, to balance dietary nutrition, to
relieve psychological symptoms, and to detect rare diseases. With the
understanding of IoT in healthcare and Health Recommender Systems, the IoT-
based Health Recommender Systems framework is developed in this study for
improving responsiveness in identifying the health problem on a daily basis and
treating the health-related issues rapidly and conveniently. Therefore, the
suggested framework aims to help people to monitor their health-related issues
so that they can prevent any health emergencies and live their healthy life.
Moreover, the integration of IoT and Health Recommender Systems ultimately
67
can provide affordable health treatment with commercialization and
popularization.
The goal of this study is accomplished by investigating healthcare services
and healthcare applications using IoT so that we know how to manage individual
health through the IoT. In addition, this project suggests the framework of IoT
and Health Recommender Systems to elaborate that the framework can provide
more advanced health treatment. Therefore, it can enhance in five aspects which
are affordability, convenience, responsiveness, treatment, and wellness of
humans.
Commercializing the integration of the IoT and Health Recommender
Systems can result in the reduction of health treatment costs by monitoring
individual health on a daily basis and it can also recommend medical advice
through doctors and healthcare professionals. Among the many IoT,
Smartwatches can be the medium of the popularization of the suggested
framework by developing the medical sensors in the Smartwatch due to its
popularity and convenient feature to wear. The IoT devices are widely used in
human life which lead to the fast responsiveness in health emergencies and
provide quick treatment using Health Recommender Systems. IoT in healthcare
can be the next innovation in people’s flourishing life and even utilization of the
Health Recommender Systems can induce the better version of healthy human
life by monitoring personal health in case of pandemic and can provide
68
recommendations for a healthy lifestyle recommendation, nutrition
recommendations and for general health information.
Future Work
While implementing this project, there are limitations of building up a
framework and implementing a case study. If relevant studies are published more
frequently, it would further support studies regarding the integration of IoT and
Health Recommender Systems. This project did not specifically focus on disease
since this project mainly focuses on the coupling of IoT and Health
Recommender Systems. The integration of IoT and Health Recommender
Systems highlights how each particular disease can be a new direction of
research for the future. Also, implementation with the algorithms of Health
Recommender Systems using specific IoT devices can be compared in future
studies.
69
APPENDIX A:
PYTHON CODE FOR CASE STUDY
70
PYTHON CODE FOR CASE STUDY
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
from collections import Counter
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.metrics import
confusion_matrix,accuracy_score,roc_curve,classification_report
from sklearn.linear_model import LogisticRegression
from sklearn.naive_bayes import GaussianNB
from sklearn.ensemble import RandomForestClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from xgboost import XGBClassifier
from mlxtend.classifier import StackingCVClassifier
data = pd.read_csv('heartdata.csv')
data.head()
data.info()
y = data["target"]
X = data.drop('target',axis=1)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.20,
random_state = 0)
print(y_test.unique())
Counter(y_train)
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)
a1 = 'Logistic Regression'
lr = LogisticRegression()
model = lr.fit(X_train, y_train)
lr_predict = lr.predict(X_test)
lr_conf_matrix = confusion_matrix(y_test, lr_predict)
lr_acc_score = accuracy_score(y_test, lr_predict)
71
print("confussion matrix")
print(lr_conf_matrix)
print("Accuracy of Logistic Regression:",lr_acc_score*100)
print(classification_report(y_test,lr_predict))
a2 = 'Naive Bayesian'
nb = GaussianNB()
nb.fit(X_train,y_train)
nbpred = nb.predict(X_test)
nb_conf_matrix = confusion_matrix(y_test, nbpred)
nb_acc_score = accuracy_score(y_test, nbpred)
print("confussion matrix")
print(nb_conf_matrix)
print("Accuracy of Naive Bayesian model:",nb_acc_score*100)
print(classification_report(y_test,nbpred))
a3 = 'Random Forest Classfier'
rf = RandomForestClassifier(n_estimators=20, random_state=12,max_depth=5)
rf.fit(X_train,y_train)
rf_predicted = rf.predict(X_test)
rf_conf_matrix = confusion_matrix(y_test, rf_predicted)
rf_acc_score = accuracy_score(y_test, rf_predicted)
print("confussion matrix")
print(rf_conf_matrix)
print("Accuracy of Random Forest:",rf_acc_score*100)
print(classification_report(y_test,rf_predicted))
a4 = 'K-NeighborsClassifier'
knn = KNeighborsClassifier(n_neighbors=10)
knn.fit(X_train, y_train)
knn_predicted = knn.predict(X_test)
knn_conf_matrix = confusion_matrix(y_test, knn_predicted)
knn_acc_score = accuracy_score(y_test, knn_predicted)
print("confussion matrix")
print(knn_conf_matrix)
print("Accuracy of K-NeighborsClassifier:",knn_acc_score*100)
print(classification_report(y_test,knn_predicted))
a5 = 'DecisionTreeClassifier'
dt = DecisionTreeClassifier(criterion = 'entropy',random_state=0,max_depth = 6)
dt.fit(X_train, y_train)
dt_predicted = dt.predict(X_test)
dt_conf_matrix = confusion_matrix(y_test, dt_predicted)
dt_acc_score = accuracy_score(y_test, dt_predicted)
72
print("confussion matrix")
print(dt_conf_matrix)
print("Accuracy of DecisionTreeClassifier:",dt_acc_score*100)
print(classification_report(y_test,dt_predicted))
a6 = 'Support Vector Classifier'
svc = SVC(kernel='rbf', C=2)
svc.fit(X_train, y_train)
svc_predicted = svc.predict(X_test)
svc_conf_matrix = confusion_matrix(y_test, svc_predicted)
svc_acc_score = accuracy_score(y_test, svc_predicted)
print("confussion matrix")
print(svc_conf_matrix)
print("Accuracy of Support Vector Classifier:",svc_acc_score*100)
print(classification_report(y_test,svc_predicted))
a7 = 'Extreme Gradient Boosting'
xgb = XGBClassifier(learning_rate=0.01, n_estimators=25,
max_depth=15,gamma=0.6, subsample=0.52,colsample_bytree=0.6,seed=27,
reg_lambda=2, booster='dart', colsample_bylevel=0.6,
colsample_bynode=0.5)
xgb.fit(X_train, y_train)
xgb_predicted = xgb.predict(X_test)
xgb_conf_matrix = confusion_matrix(y_test, xgb_predicted)
xgb_acc_score = accuracy_score(y_test, xgb_predicted)
print("confussion matrix")
print(xgb_conf_matrix)
print("Accuracy of Extreme Gradient Boosting:",xgb_acc_score*100)
print(classification_report(y_test,xgb_predicted))
%matplotlib inline
imp_feature = pd.DataFrame({'Feature': ['age', 'sex', 'cp', 'trestbps', 'chol', 'fbs',
'restecg', 'thalach',
'exang', 'oldpeak', 'slope', 'ca', 'thal'], 'Importance':
xgb.feature_importances_})
plt.figure(figsize=(12,8))
plt.title("Feature importance for heart diseases")
plt.xlabel("features")
plt.ylabel("importance")
plt.bar(imp_feature['Feature'],imp_feature['Importance'])
plt.show()
model_ev = pd.DataFrame({'Model': ['Decision Tree','Logistic Regression','Naive
Bayesian','Extreme Gradient Boosting','Random Forest',
73
'K-Nearest Neighbors','Support Vector Machine'], 'Accuracy':
[dt_acc_score*100,lr_acc_score*100,
nb_acc_score*100,xgb_acc_score*100,rf_acc_score*100,knn_acc_score*100,sv
c_acc_score*100]})
model_ev
%matplotlib inline
plt.figure(figsize=(12,8))
plt.title("Accuracy of different Machine Learning algorithms")
plt.xlabel("Accuracy %")
plt.ylabel("Algorithms")
plt.barh(model_ev['Model'],model_ev['Accuracy'])
plt.show()
corr=data.corr()
corr
data = data.drop(['target'],axis =1)
plt.figure(figsize=(10,10))
sns.heatmap(X.corr(), vmax=1, center=0, square=True, linewidths=.5,
cbar_kws={"shrink": .6}, annot=True, cmap='Blues')
plt.tight_layout()
plt.show()
74
REFERENCES
Aamodt, A., & Enric, P. (1994). Case-based reasoning: Foundational issues,
methodological variations, and system approaches. AI Communications,
7(1), 39–59. https://doi.org/10.3233/aic-1994-7104
Abdmeziem, M. R., Tandjaoui, D., & Romdhani, I. (2015). Architecting the
Internet of Things: State of the Art. In Robots and sensor clouds (pp. 55–
75). essay, Springer.
Adomavicius, G., & Tuzhilin, A. (2005). Toward the next generation of
Recommender Systems: A survey of the state-of-the-art and possible
extensions. IEEE Transactions on Knowledge and Data Engineering,
17(6), 734–749. https://doi.org/10.1109/tkde.2005.99
Agustine, L., Muljono, I., Angka, P. R., Gunadhi, A., Lestariningsih, D., &
Weliamto, W. A. (2018). Heart rate monitoring device for arrhythmia using
pulse oximeter sensor based on Android. 2018 International Conference
on Computer Engineering, Network and Intelligent Multimedia (CENIM),
106–111. https://doi.org/10.1109/cenim.2018.8711120
Ahmad, E. (2020). Meezaj: An interactive system for real-time mood
measurement and reflection based on internet of things. International
Journal of Advanced Computer Science and Applications, 11(11), 629–
636. https://doi.org/10.14569/ijacsa.2020.0111177
Ahmed, M. I., & Kannan, G. (2021). Secure and lightweight privacy preserving
internet of things integration for Remote Patient Monitoring. Journal of
75
King Saud University - Computer and Information Sciences.
https://doi.org/10.1016/j.jksuci.2021.07.016
Akhbarifar, S., Javadi, H. H. S., Rahmani, A. M., & Hosseinzadeh, M. (2020). A
secure remote health monitoring model for early disease diagnosis in
cloud-based IOT environment. Personal and Ubiquitous Computing.
https://doi.org/10.1007/s00779-020-01475-3
Akmandor, A. O., & Jha, N. K. (2017). Smart health care: An edge-side
computing perspective. IEEE Consumer Electronics Magazine, 7(1), 29–
37. https://doi.org/10.1109/mce.2017.2746096
Alam, M. G. R., Abedin, S. F., Moon, S. I., & Hong, C. S. (2017). CNN based
Mood Mining through IoT-based Physiological Sensors Observation.
한국컴퓨터종합학술대회
논문집
, 1301–1303.
Alarcón-Paredes, A., Francisco-García, V., Guzmán-Guzmán, I. P., Cantillo-
Negrete, J., Cuevas-Valencia, R. E., & Alonso-Silverio, G. A. (2019). An
IOT-based non-invasive glucose level monitoring system using Raspberry
Pi. Applied Sciences, 9(15), 3046. https://doi.org/10.3390/app9153046
Al-Kababji, A., Shidqi, L., Boukhennoufa, I., Amira, A., Bensaali, F., Gastli, M. S.,
Jarouf, A., Aboueata, W., & Abdalla, A. (2019). IOT-based fall and ECG
Monitoring System: Wireless Communication System Based Firebase
realtime database. 2019 IEEE SmartWorld, Ubiquitous Intelligence &
Computing, Advanced & Trusted Computing, Scalable Computing &
Communications, Cloud & Big Data Computing, Internet of People and
76
Smart City Innovation
(SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI), 1480–1485.
https://doi.org/10.1109/smartworld-uic-atc-scalcom-iop-sci.2019.00267
Almeida, J. R., Monteiro, E., Silva, L. B., Pazos Sierra, A., & Oliveira, J. L.
(2020). A recommender system to help discovering cohorts in rare
diseases. 2020 IEEE 33rd International Symposium on Computer-Based
Medical Systems (CBMS). https://doi.org/10.1109/cbms49503.2020.00012
American Cancer Society. (2019, May 2). Medical costs create hardships for
more than half of Americans. ScienceDaily. Retrieved November 3, 2021,
from https://www.sciencedaily.com/releases/2019/05/190502100818.htm.
Amin, S. U., Hossain, M. S., Muhammad, G., Alhussein, M., & Rahman, M. A.
(2019). Cognitive smart healthcare for pathology detection and monitoring.
IEEE Access, 7, 10745–10753.
https://doi.org/10.1109/access.2019.2891390
Ansari, D. B., Rehman, A. U., & Mughal, R. A. (2018). Internet of things (IOT)
protocols: A brief exploration of mqtt and Coap. International Journal of
Computer Applications, 179(27), 9–14.
https://doi.org/10.5120/ijca2018916438
Apple. (2021). Apple. Official Apple Support. Retrieved November 5, 2021, from
https://support.apple.com/kb/SP860?viewlocale=en_US.
Apple. (2021). Healthcare - Apple Watch. Apple. Retrieved November 5, 2021,
from http://www.apple.com/healthcare/apple-watch/.
77
Archenaa, J., & Anita, E. A. M. (2017). Health Recommender System using Big
data analytics. Journal of Management Science and Business Intelligence,
2, 17–24. https://doi.org/10.5281/zenodo. 10.5281/zenodo.835606
Balabanović, M., & Shoham, Y. (1997). Fab: content-based, collaborative
recommendation. Communications of the ACM, 40(3), 66–72.
https://doi.org/10.1145/245108.245124
Banerjee, S. S., Hemphill, T., & Longstreet, P. (2017). Wearable devices and
healthcare: Data Sharing and Privacy. The Information Society, 34(1), 49–
57. https://doi.org/10.1080/01972243.2017.1391912
Bansal, M., & Gandhi, B. (2019). IOT & Big Data in Smart Healthcare (ECG
monitoring). 2019 International Conference on Machine Learning, Big
Data, Cloud and Parallel Computing (COMITCon), 390–396.
https://doi.org/10.1109/comitcon.2019.8862197
Behera, R. K., Bala, P. K., & Dhir, A. (2019). The emerging role of Cognitive
Computing in healthcare: A systematic literature review. International
Journal of Medical Informatics, 129, 154–166.
https://doi.org/10.1016/j.ijmedinf.2019.04.024
Bharadwaj, S. A., Yarravarapu, D., Reddy, S. C. K., Prudhvi, T., Sandeep, K. S.
P., & Reddy, O. S. D. (2017). Enhancing healthcare using M-care box
(monitoring non-compliance of medication). 2017 International Conference
on Innovative Mechanisms for Industry Applications (ICIMIA), 167–171.
https://doi.org/10.1109/icimia.2017.7975594
78
Bobadilla, J., Ortega, F., Hernando, A., & Gutiérrez, A. (2013). Recommender
Systems survey. Knowledge-Based Systems, 46, 109–132.
https://doi.org/10.1016/j.knosys.2013.03.012
Briseno, M. V., Cota, C. N.-, Hipolito, J. I. N.-, Garcia, E. J., & Lopez, J. D. S.
(2012). A proposal for using the internet of things concept to increase
children's health awareness. CONIELECOMP 2012, 22nd International
Conference on Electrical Communications and Computers.
https://doi.org/10.1109/conielecomp.2012.6189903
Burke, R. (2002). Hybrid Recommender Systems: Survey and Experiments. User
Modeling and User-Adapted Interaction, 331–370.
https://doi.org/10.1023/A:1021240730564
Chang, D., Gui, H., Fan, R., Fan, Z., & Tian, J. (2019). Application of improved
collaborative filtering in the recommendation of e-commerce commodities.
International Journal of Computers Communications & Control, 14(4),
489–502. https://doi.org/10.15837/ijccc.2019.4.3594
Chao, P. C.-P., & Tu, T.-Y. (2017). Using the time-domain characterization for
estimation continuous blood pressure via neural network method. ASME
2017 Conference on Information Storage and Processing Systems.
https://doi.org/10.1115/isps2017-5471
Chavan, P., Thoms, B., & Isaacs, J. (2021). A recommender system for Healthy
Food Choices: Building a hybrid model for recipe recommendations using
79
Big Data Sets. Proceedings of the 54th Hawaii International Conference
on System Sciences. https://doi.org/10.24251/hicss.2021.458
Chuquimarca, L., Roca, D., Torres, W., Amaya, L., Orozco, J., & Sanchez, D.
(2020). Mobile IOT device for BPM monitoring people with heart problems.
2020 International Conference on Electrical, Communication, and
Computer Engineering (ICECCE).
https://doi.org/10.1109/icecce49384.2020.9179293
Codina-Filbà, J., Escalera, S., Escudero, J., Antens, C., Buch-Cardona, P., &
Farrús, M. (2021). Mobile ehealth platform for home monitoring of bipolar
disorder. MultiMedia Modeling, 330–341. https://doi.org/10.1007/978-3-
030-67835-7_28
Čolaković, A., & Hadžialić, M. (2018). Internet of things (IOT): A review of
Enabling Technologies, challenges, and open research issues. Computer
Networks, 144, 17–39. https://doi.org/10.1016/j.comnet.2018.07.017
Coleman, J. J., & Pontefract, S. K. (2016). Adverse drug reactions. Clinical
Medicine, 16(5), 481–485. https://doi.org/10.7861/clinmedicine.16-5-481
Czeisler, M. É., Marynak, K., Clarke, K. E. N., Salah, Z., Shakya, I., Thierry, J. A.
M., Ali, N., McMillan, H., Wiley, J. F., Weaver, M. D., Czeisler, C. A.,
Rajaratnam, S. M. W., & Howard, M. E. (2020, September 10). Delay or
avoidance of medical care because of COVID-19–related concerns -
United States, June 2020. Centers for Disease Control and Prevention.
80
Retrieved October 6, 2021, from
https://www.cdc.gov/mmwr/volumes/69/wr/mm6936a4.htm.
Deshpande, M., & Karypis, G. (2004). Item-based top-n recommendation
algorithms. ACM Transactions on Information Systems, 22(1), 143–177.
https://doi.org/10.1145/963770.963776
Dickson, E. L., Ding, E. Y., Saczynski, J. S., Han, D., Moonis, M., Fitzgibbons, T.
P., Barton, B., Chon, K., & McManus, D. D. (2021). Smartwatch
monitoring for atrial fibrillation after stroke—the Pulsewatch study:
Protocol for a multiphase randomized controlled trial. Cardiovascular
Digital Health Journal, 2(4), 231–241.
https://doi.org/10.1016/j.cvdhj.2021.07.002
Dinh, A., Luu, L., & Cao, T. (2017). Blood pressure measurement using finger
ECG and photoplethysmogram for IOT. 6th International Conference on
the Development of Biomedical Engineering in Vietnam (BME6), 83–89.
https://doi.org/10.1007/978-981-10-4361-1_14
Djelouat, H., Disi, M. A., Boukhenoufa, I., Amira, A., Bensaali, F., Kotronis, C.,
Politi, E., Nikolaidou, M., & Dimitrakopoulos, G. (2020). Real-time ECG
monitoring using compressive sensing on a heterogeneous multicore
edge-device. Microprocessors and Microsystems, 72, 102839.
https://doi.org/10.1016/j.micpro.2019.06.009
Drew, B. J., Califf, R. M., Funk, M., Kaufman, E. S., Krucoff, M. W., Laks, M. M.,
Macfarlane, P. W., Sommargren, C., Swiryn, S., & Van Hare, G. F. (2004).
81
Practice standards for electrocardiographic monitoring in hospital settings.
Circulation, 110(17), 2721–2746.
https://doi.org/10.1161/01.cir.0000145144.56673.59
Dutta, M. J., Kaur-Gill, S., Tan, N., & Lam, C. (2017). MHealth, health, and
mobility: A culture-centered interrogation. Mobile Communication in Asia:
Local Insights, Global Implications, 91–107. https://doi.org/10.1007/978-
94-024-1251-2_6
Erdeniz, S. P., Menychtas, A., Maglogiannis, I., Felfernig, A., & Tran, T. N.
(2019). Recommender systems for IOT enabled quantified-self
applications. Evolving Systems, 11(2), 291–304.
https://doi.org/10.1007/s12530-019-09302-8
Ertuğrul, D. Ç., & Elçi, A. (2019). A survey on semanticized and Personalized
Health Recommender Systems. Expert Systems, 37(4).
https://doi.org/10.1111/exsy.12519
Fan, J., Wang, X., Wu, L., Zhou, H., Zhang, F., Yu, X., Lu, X., & Xiang, Y. (2018).
Comparison of support vector machine and extreme gradient boosting for
predicting daily global solar radiation using temperature and precipitation
in humid subtropical climates: A case study in China. Energy Conversion
and Management, 164, 102–111.
https://doi.org/10.1016/j.enconman.2018.02.087
Fayyaz, Z., Ebrahimian, M., Nawara, D., Ibrahim, A., & Kashef, R. (2020).
Recommendation systems: Algorithms, challenges, metrics, and business
82
opportunities. Applied Sciences, 10(21), 7748.
https://doi.org/10.3390/app10217748
Felfernig, A., & Burke, R. (2008). Constraint-based Recommender Systems:
technologies and research issues. Proceedings of the 10th International
Conference on Electronic Commerce - ICEC '08, 1–10.
https://doi.org/10.1145/1409540.1409544
Felfernig, A., Friedrich, G., Jannach, D., & Zanker, M. (2011). Developing
Constraint-based Recommenders. In Recommender Systems Handbook
(pp. 187–215). essay, Springer.
Fu, Y., & Liu, J. (2015). System design for wearable blood oxygen saturation and
Pulse Measurement Device. Procedia Manufacturing, 3, 1187–1194.
https://doi.org/10.1016/j.promfg.2015.07.197
Garg, U., Ghanshala, K. K., Joshi, R. C., & Chauhan, R. (2018). Design and
implementation of smart wheelchair for quadriplegia patients using IOT.
2018 First International Conference on Secure Cyber Computing and
Communication (ICSCCC), 106–110.
https://doi.org/10.1109/icsccc.2018.8703354
Ghazal, T. M., Hasan, M. K., Alshurideh, M. T., Alzoubi, H. M., Ahmad, M.,
Akbar, S. S., Al Kurdi, B., & Akour, I. A. (2021). IOT for Smart Cities:
Machine Learning Approaches in smart healthcare—A Review. Future
Internet, 13(8), 218. https://doi.org/10.3390/fi13080218
83
Ghazanfar, M., & Bennett, A. P. (2010). An improved switching hybrid
recommender system using naive Bayes classifier and collaborative
filtering. The 2010 IAENG International Conference on Data Mining and
Applications. https://doi.org/eprints.soton.ac.uk/268483/
Ghorbel, A., Bouguerra, S., Amor, N. B., & Jallouli, M. (2018). Cloud based
mobile application for remote control of intelligent wheelchair. 2018 14th
International Wireless Communications & Mobile Computing Conference
(IWCMC), 1249–1254. https://doi.org/10.1109/iwcmc.2018.8450366
Gonçalo, M. (2020). Patch Testing in Adverse Drug Reactions. In D. P.
Bruynzeel (Ed.), Contact Dermatitis (pp. 571–591). essay, Springer,
Cham.
Guk, K., Han, G., Lim, J., Jeong, K., Kang, T., Lim, E.-K., & Jung, J. (2019).
Evolution of wearable devices with real-time disease monitoring for
Personalized Healthcare. Nanomaterials, 9(6), 813.
https://doi.org/10.3390/nano9060813
Gunawan, I. P., Andayani, D. H., Triwiyanto, T., Yulianto, E., Rahmawati, T.,
Soetjiatie, L., & Musvika, S. D. (2020). Design and development of
telemedicine based heartbeat and Body Temperature Monitoring Tools.
IOP Conference Series: Materials Science and Engineering, 850, 012018.
https://doi.org/10.1088/1757-899x/850/1/012018
Gundu, S. (2020). A novel IOT based solution for monitoring and alerting
bronchial asthma patients. International Journal of Research in
84
Engineering, Science and Management, 3(10), 120–123.
https://doi.org/10.47607/ijresm.2020.348
Guntha, R. (2019). IOT architectures for noninvasive blood glucose and Blood
Pressure Monitoring. 2019 9th International Symposium on Embedded
Computing and System Design (ISED).
https://doi.org/10.1109/ised48680.2019.9096233
Gyrard, A., & Sheth, A. (2020). Iamhappy: Towards an IOT knowledge-based
cross-domain well-being recommendation system for Everyday happiness.
Smart Health, 15, 1–15. https://doi.org/10.1016/j.smhl.2019.100083
Hemanth, D. J., Anitha, J., & Tsihrintzis, G. A. (2021). Internet of medical things:
Remote healthcare systems and applications. Springer.
Hors-Fraile, S., Benjumea, F. J. N., Hernández, L. C., Ruiz, F. O., & Fernandez-
Luque, L. (2016). Design of two combined health recommender systems
for tailoring messages in a smoking cessation app. Cornell University, 1.
https://doi.org/arXiv:1608.07192
Hui, C. Y., McKinstry, B., Fulton, O., Buchner, M., & Pinnock, H. (2021). Patients’
and clinicians’ visions of a future internet-of-things system to support
asthma self-management: Mixed Methods Study. Journal of Medical
Internet Research, 23(4). https://doi.org/10.2196/22432
IBM Cloud Education. (2020, May 27). AI vs. Machine Learning vs. Deep
Learning vs. neural networks: What's the difference? IBM. Retrieved
85
November 8, 2021, from https://www.ibm.com/cloud/blog/ai-vs-machine-
learning-vs-deep-learning-vs-neural-networks.
Ibrahim, I. M., & Abdulazeez, A. M. (2021). The role of machine learning
algorithms for diagnosing diseases. Journal of Applied Science and
Technology Trends, 2(01), 10–19. https://doi.org/10.38094/jastt20179
Isakadze, N., & Martin, S. S. (2020). How useful is the smartwatch ECG? Trends
in Cardiovascular Medicine, 30(7), 442–448.
https://doi.org/10.1016/j.tcm.2019.10.010
Islam, S. M. R., Kwak, D., Kabir, M. H., Hossain, M., & Kwak, K.-S. (2015). The
internet of things for health care: A comprehensive survey. IEEE Access,
3, 678–708. https://doi.org/10.1109/access.2015.2437951
Istepanian, R. S. H., Casiglia, D., & Gregory, J. W. (2017). Mobile Health (M-
health) for diabetes management. British Journal of Healthcare
Management, 23(3), 102–108.
https://doi.org/10.12968/bjhc.2017.23.3.102
Istepanian, R. S. H., Hu, S., Philip, N. Y., & Sungoor, A. (2011). The potential of
Internet of m-health Things “m-IoT” for non-invasive glucose level sensing.
2011 Annual International Conference of the IEEE Engineering in
Medicine and Biology Society.
https://doi.org/10.1109/iembs.2011.6091302
Iwendi, C., Khan, S., Anajemba, J. H., Bashir, A. K., & Noor, F. (2020). Realizing
an efficient IOMT-assisted patient diet recommendation system through
86
Machine Learning Model. IEEE Access, 8, 28462–28474.
https://doi.org/10.1109/access.2020.2968537
Jagadeeswari, V., Subramaniyaswamy, V., Logesh, R., & Vijayakumar, V.
(2018). A study on medical internet of things and Big Data in personalized
healthcare system. Health Information Science and Systems, 6(1).
https://doi.org/10.1007/s13755-018-0049-x
Janosi, A., Steinbrunn, W., Pfisterer, M., Detrano, R., & M.D., M. D. (1988, July
1). Heart Disease. UCI Machine Learning Repository: Heart disease data
set. Retrieved November 3, 2021, from
https://archive.ics.uci.edu/ml/datasets/heart+disease.
Jara, A. J., Belchi, F. J., Alcolea, A. F., Santa, J., Zamora-Izquierdo, M. A., &
Gomez-Skarmeta, A. F. (2010). A pharmaceutical intelligent information
system to detect allergies and adverse drugs reactions based on internet
of things. 2010 8th IEEE International Conference on Pervasive
Computing and Communications Workshops (PERCOM Workshops).
https://doi.org/10.1109/percomw.2010.5470547
Kamišalić, A., Fister, I., Turkanović, M., & Karakatič, S. (2018). Sensors and
functionalities of non-invasive wrist-wearable devices: A Review. Sensors,
18(6), 1714. https://doi.org/10.3390/s18061714
Kao, H.-T., Yan, S., Hosseinmardi, H., Narayanan, S., Lerman, K., & Ferrara, E.
(2020). User-based collaborative filtering Mobile Health System.
87
Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous
Technologies, 4(4), 1–17. https://doi.org/10.1145/3432703
Karagiannis, D., & Nikita, K. S. (2020). Design and development of a 3D printed
IOT portable pillbox for continuous medication adherence. 2020 IEEE
International Conference on Smart Internet of Things (SmartIoT), 352–
353. https://doi.org/10.1109/smartiot49966.2020.00066
Kashou, A. H., Basit, H., & Malik, A. (2021). ST Segment. In StatPearls. essay,
Treasure Island (FL): StatPearls.
Kaur, J., & Mann, K. S. (2017). AI based healthcare platform for real time,
predictive and prescriptive analytics using reactive programming. Journal
of Physics: Conference Series, 933, 012010. https://doi.org/10.1088/1742-
6596/933/1/012010
Kelati, A., Dhaou, I. B., & Tenhunen, H. (2018). Biosignal Monitoring Platform
Using Wearable IoT. PROCEEDING OF THE 22ND CONFERENCE OF
FRUCT ASSOCIATION, 332–337.
https://doi.org/fruct.org/publications/acm22/files/Kel.pdf
Koren, Y., Bell, R., & Volinsky, C. (2009). Matrix factorization techniques for
Recommender Systems. Computer, 42(8), 30–37.
https://doi.org/10.1109/mc.2009.263
Kumar, D., Malhotra, R., & Sharma, S. R. (2020). Design and construction of a
smart wheelchair. Procedia Computer Science, 172, 302–307.
https://doi.org/10.1016/j.procs.2020.05.048
88
Kumar, G. (2020). A Review On Machine Learning Concept And Its Algorithms.
Alochana Chakra Journal, 9534–9544.
https://doi.org/www.parishodhpu.com/gallery/1164-p-march-2020.pdf
Kumar, M. A., Vimala, R., & Britto, K. R. A. (2019). A cognitive technology based
healthcare monitoring system and medical data transmission.
Measurement, 146, 322–332.
https://doi.org/10.1016/j.measurement.2019.03.017
Kumar, P. M., & Gandhi, U. D. (2018). A novel three-tier internet of things
architecture with machine learning algorithm for early detection of heart
diseases. Computers & Electrical Engineering, 65, 222–235.
https://doi.org/10.1016/j.compeleceng.2017.09.001
Lee, Y. K., Lim, J. M., Eu, K. S., Goh, Y. H., & Tew, Y. (2017). Real time image
processing based obstacle avoidance and navigation system for
autonomous wheelchair application. 2017 Asia-Pacific Signal and
Information Processing Association Annual Summit and Conference
(APSIPA ASC), 380–385. https://doi.org/10.1109/apsipa.2017.8282062
Leonhardt, M. (2020, March 12). Nearly 1 in 4 Americans are skipping medical
care because of the cost. CNBC. Retrieved October 6, 2021, from
https://www.cnbc.com/2020/03/11/nearly-1-in-4-americans-are-skipping-
medical-care-because-of-the-cost.html.
Lim, S. (2021, May 27). Global smartwatch shipments jump 35% yoy in Q1 2021.
Counterpoint Research. Retrieved October 25, 2021, from
89
https://www.counterpointresearch.com/global-smartwatch-shipments-q1-
2021/.
Liu, B., Xiong, H., Papadimitriou, S., Fu, Y., & Yao, Z. (2015). A general
geographical probabilistic factor model for point of interest
recommendation. IEEE Transactions on Knowledge and Data
Engineering, 27(5), 1167–1179.
https://doi.org/10.1109/tkde.2014.2362525
Liu, M.-L. (2012). Internet of Things-based electrocardiogram monitoring system.
Chinese Patent, 102, 118.
Lombardi, M., Pascale, F., & Santaniello, D. (2021). Internet of things: A general
overview between architectures, protocols and applications. Information,
12(2), 87. https://doi.org/10.3390/info12020087
Lu, S., Wang, B., Wang, H., & Hong, Q. (2018). A hybrid collaborative filtering
algorithm based on KNN and gradient boosting. 2018 13th International
Conference on Computer Science & Education (ICCSE), 432–436.
https://doi.org/10.1109/iccse.2018.8468751
Luo, X., Zhou, M. C., Li, S., You, Z., Xia, Y., & Zhu, Q. (2016). A nonnegative
latent factor model for large-scale sparse matrices in recommender
systems via alternating direction method. IEEE Transactions on Neural
Networks and Learning Systems, 27(3), 579–592.
https://doi.org/10.1109/tnnls.2015.2415257
90
Madakam, S., Ramaswamy, R., & Tripathi, S. (2015). Internet of things (IOT): A
literature review. Journal of Computer and Communications, 03(05), 164–
173. https://doi.org/10.4236/jcc.2015.35021
Mahesh, B. (2020). Machine Learning Algorithms - A Review. International
Journal of Science and Research (IJSR), 9(1).
https://doi.org/10.21275/ART20203995
Maranda, A. P., Kaczmarek, D., Kryvinska, N., & Xhafa, F. (2018). Studying
usability of AI in the IOT systems/paradigm through embedding NN
techniques into Mobile Smart Service System. Computing, 101(11), 1661–
1685. https://doi.org/10.1007/s00607-018-0680-z
Meshref, H. (2019). Cardiovascular disease diagnosis: A machine learning
interpretation approach. International Journal of Advanced Computer
Science and Applications, 10(12), 258–269.
https://doi.org/10.14569/ijacsa.2019.0101236
Nakhla, Z., Nouira, K., & Ferchichi, A. (2018). Prescription Adverse Drug Events
System (prescade) based on ontology and internet of things. The
Computer Journal, 62(6), 801–805. https://doi.org/10.1093/comjnl/bxy076
Nasrullah, P. (2021, September 21). Internet of things in Healthcare:
Applications, benefits, and challenges. Peerbits. Retrieved October 6,
2021, from https://www.peerbits.com/blog/internet-of-things-healthcare-
applications-benefits-and-challenges.html.
91
Nave, C., & Postolache, O. (2018). Smart Walker based IOT physical
rehabilitation system. 2018 International Symposium in Sensing and
Instrumentation in IoT Era (ISSI).
https://doi.org/10.1109/issi.2018.8538210
Nawara, D., & Kashef, R. (2020). IOT-based recommendation systems – an
overview. 2020 IEEE International IOT, Electronics and Mechatronics
Conference (IEMTRONICS).
https://doi.org/10.1109/iemtronics51293.2020.9216391
Ngu, A. H., Gutierrez, M., Metsis, V., Nepal, S., & Sheng, Q. Z. (2016). IOT
middleware: A survey on issues and Enabling Technologies. IEEE Internet
of Things Journal, 4(1), 1–1. https://doi.org/10.1109/jiot.2016.2615180
Nilashi, M., Ibrahim, O. B., Ithnin, N., & Zakaria, R. (2014). A multi-criteria
recommendation system using dimensionality reduction and neuro-fuzzy
techniques. Soft Computing, 19(11), 3173–3207.
https://doi.org/10.1007/s00500-014-1475-6
Oracle. (n.d.). What is the internet of things (IOT)? Oracle. Retrieved October 6,
2021, from https://www.oracle.com/internet-of-things/what-is-iot/.
Ota, H., Chao, M., Gao, Y., Wu, E., Tai, L.-C., Chen, K., Matsuoka, Y., Iwai, K.,
Fahad, H. M., Gao, W., Nyein, H. Y., Lin, L., & Javey, A. (2017). 3D
printed “Earable” smart devices for real-time detection of core body
temperature. ACS Sensors, 2(7), 990–997.
https://doi.org/10.1021/acssensors.7b00247
92
Pandey, P. S. (2017). Machine learning and IOT for prediction and detection of
stress. 2017 17th International Conference on Computational Science and
Its Applications (ICCSA). https://doi.org/10.1109/iccsa.2017.8000018
Park, Y., Park, S., Jung, W., & Lee, S.-goo. (2015). Reversed CF: A fast
collaborative filtering algorithm using a K-nearest neighbor graph. Expert
Systems with Applications, 42(8), 4022–4028.
https://doi.org/10.1016/j.eswa.2015.01.001
Pennic, F. (2019, April 15). 2 in 3 physicians plan to use remote patient
monitoring in the future. Healthcare IT News. Retrieved October 6, 2021,
from https://hitconsultant.net/2019/04/15/cta-remote-patient-monitoring-
survey/#.YVUGCprMJPZ.
Piwek, L., Ellis, D. A., Andrews, S., & Joinson, A. (2016). The Rise of Consumer
Health Wearables: Promises and barriers. PLOS Medicine, 13(2).
https://doi.org/10.1371/journal.pmed.1001953
Polaris Market Research. (2020, July). Wearable Medical Device Market Size:
Global Industry Report, 2027. Polaris Market Research: Global Market
Research Reports and Consulting. Retrieved November 1, 2021, from
https://www.polarismarketresearch.com/industry-analysis/wearable-
medical-devices-market.
Powell, B. D., Saxon, L. A., Boehmer, J. P., Day, J. D., Gilliam, F. R.,
Heidenreich, P. A., Jones, P. W., Rousseau, M. J., & Hayes, D. L. (2013).
Survival after shock therapy in implantable cardioverter-defibrillator and
93
cardiac resynchronization therapy-defibrillator recipients according to
Rhythm shocked. Journal of the American College of Cardiology, 62(18),
1674–1679. https://doi.org/10.1016/j.jacc.2013.04.083
Pradhan, B., Bhattacharyya, S., & Pal, K. (2021). IOT-based applications in
healthcare devices. Journal of Healthcare Engineering, 2021, 1–18.
https://doi.org/10.1155/2021/6632599
Prasad, A. K. (2020). Smart asthma alert using IOT and predicting threshold
values using decision tree classifier. Applications of Internet of Things,
137, 141–150. https://doi.org/10.1007/978-981-15-6198-6_13
Qi, J., Yang, P., Waraich, A., Deng, Z., Zhao, Y., & Yang, Y. (2018). Examining
sensor-based physical activity recognition and monitoring for healthcare
using internet of things: A systematic review. Journal of Biomedical
Informatics, 87, 138–153. https://doi.org/10.1016/j.jbi.2018.09.002
Raji, A., Devi, P. K., Jeyaseeli, P. G., & Balaganesh, N. (2016). Respiratory
monitoring system for asthma patients based on IOT. 2016 Online
International Conference on Green Engineering and Technologies (IC-
GET), 1–6. https://doi.org/10.1109/get.2016.7916737
Razdan, S., & Sharma, S. (2021). Internet of medical things (IOMT): Overview,
emerging technologies, and case studies. IETE Technical Review, 1–14.
https://doi.org/10.1080/02564602.2021.1927863
Ricci, F., Rokach, L., Shapira, B., & Kantor, P. B. (2011). Recommender
Systems Handbook. Springer.
94
Ruiz, M. N. S., García, J. M., & Fernández, B. M. (2009). Body temperature and
its importance as a vital constant. Revista Enfermeria, 32(9), 44–52.
https://doi.org/pubmed.ncbi.nlm.nih.gov/19839238/#affiliation-1
Saad, L. (2021, August 13). More Americans delaying medical treatment due to
cost. Gallup. Retrieved November 3, 2021, from
https://news.gallup.com/poll/269138/americans-delaying-medical-
treatment-due-cost.aspx.
Sahoo, A. K., Pradhan, C., Barik, R. K., & Dubey, H. (2019). DeepReco: Deep
Learning Based Health Recommender System using collaborative filtering.
Computation, 7(2), 25. https://doi.org/10.3390/computation7020025
Sahu, M. lata, Atulkar, M., & Ahirwal, M. K. (2020). Comprehensive investigation
on IOT based Smart Healthcare System. 2020 First International
Conference on Power, Control and Computing Technologies (ICPC2T).
https://doi.org/10.1109/icpc2t48082.2020.9071442
Sainato, M. (2020, January 7). The Americans dying because they can't afford
medical care. The Guardian. Retrieved October 6, 2021, from
https://www.theguardian.com/us-news/2020/jan/07/americans-healthcare-
medical-costs.
Salton, G., Wong, A., & Yang, C. (1975). A vector space model for automatic
indexing. Communications of the ACM, 18(11), 613–620.
https://doi.org/10.1145/361219.361220
95
Sana, F., Isselbacher, E. M., Singh, J. P., Heist, E. K., Pathik, B., & Armoundas,
A. A. (2020). Wearable devices for ambulatory cardiac monitoring. Journal
of the American College of Cardiology, 75(13), 1582–1592.
https://doi.org/10.1016/j.jacc.2020.01.046
Samsung. (2021, August). Galaxy Watch4 Smartwatch: Samsung Us. Samsung
Electronics America. Retrieved October 25, 2021, from
https://www.samsung.com/us/watches/galaxy-watch4/.
Sandeepa, C., Moremada, C., Dissanayaka, N., Gamage, T., & Liyanage, M.
(2020). An emergency situation detection system for Ambient Assisted
Living. 2020 IEEE International Conference on Communications
Workshops (ICC Workshops).
https://doi.org/10.1109/iccworkshops49005.2020.9145053
Satamraju, K. P., & B, M. (2020). Proof of concept of scalable integration of
internet of things and blockchain in Healthcare. Sensors, 20(5), 1389.
https://doi.org/10.3390/s20051389
Schafer, J. B., Frankowski, D., Herlocker, J., & Sen, S. (2007). Collaborative
Filtering Recommender Systems. In The adaptive web methods and
strategies of web personalization (pp. 291–324). essay, Springer.
Schurrer, J., O’Malley, A., Wilson, C., McCall, N., & Jain, N. (2017). Evaluation of
the Diffusion and Impact of the Chronic Care Management (CCM)
Services: Final Report. Mathematica Policy Research.
https://doi.org/https://www.mathematica.org/publications/evaluation-of-the-
96
diffusion-and-impact-of-the-chronic-care-management-ccm-services-final-
report
Sebastiani, F. (2002). Machine learning in automated text categorization. ACM
Computing Surveys, 34(1), 1–47. https://doi.org/10.1145/505282.505283
Shah, S., Badshah, F., Dad, F., Amin, N., & Jan, M. (2018). Cloud-Assisted IoT-
Based Smart Respiratory Monitoring System for Asthma Patients. In
Applications of intelligent technologies in Healthcare (pp. 77`-86). essay,
Springer International Publishing.
Shahamabadi, M. S., Ali, B. B. M., Varahram, P., & Jara, A. J. (2013). A network
mobility solution based on 6LoWPAN Hospital Wireless Sensor Network
(Nemo-HWSN). 2013 Seventh International Conference on Innovative
Mobile and Internet Services in Ubiquitous Computing.
https://doi.org/10.1109/imis.2013.157
Shardanand, U., & Maes, P. (1995). Social Information Filtering: algorithms for
automating “word of mouth.” Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems - CHI '95, 210–217.
https://doi.org/10.1145/223904.223931
Sharma, C., & Gondhi, N. K. (2018). Communication protocol stack for
constrained IOT Systems. 2018 3rd International Conference On Internet
of Things: Smart Innovation and Usages (IoT-SIU).
https://doi.org/10.1109/iot-siu.2018.8519904
97
Singh, K., Kaushik, K., Ahatsham, & Shahare, V. (2020). Role and Impact of
Wearables in IoT Healthcare. In Proceedings of the third international
conference on computational intelligence and informatics: ICCII 2018 (Vol.
1090, pp. 735–742). essay, Springer Singapore.
Slotwiner, D., Varma, N., Akar, J. G., Annas, G., Beardsall, M., Fogel, R. I.,
Galizio, N. O., Glotzer, T. V., Leahy, R. A., Love, C. J., McLean, R. C.,
Mittal, S., Morichelli, L., Patton, K. K., Raitt, M. H., Pietro Ricci, R.,
Rickard, J., Schoenfeld, M. H., Serwer, G. A., … Yu, C.-M. (2015). HRS
expert consensus statement on Remote Interrogation and monitoring for
cardiovascular implantable electronic devices. Heart Rhythm, 12(7).
https://doi.org/10.1016/j.hrthm.2015.05.008
Sutjiredjeki, E., Basjaruddin, N. C., Fajrin, D. N., & Noor, F. (2020). Development
of NFC and IOT-enabled measurement devices for improving health care
delivery of Indonesian children. Journal of Physics: Conference Series,
1450, 012072. https://doi.org/10.1088/1742-6596/1450/1/012072
Urgiriye, A., & Bhartiya, R. (2020). Review of Machine Learning Algorithm on
Cancer Data Set. International Journal of Scientific Research &
Engineering Trends, 6(6), 3259–3267. https://doi.org/ijsret.com/wp-
content/uploads/2020/11/IJSRET_V6_issue6_720.pdf
Valdiviezo-Diaz, P., Ortega, F., Cobos, E., & Lara-Cabrera, R. (2019). A
collaborative filtering approach based on naïve Bayes classifier. IEEE
Access, 7, 108581–108592. https://doi.org/10.1109/access.2019.2933048
98
Valenzuela, F., García, A., Ruiz., E., Vázquez, M., Cortez, J., & Espinoza, A.
(2020). An IOT-based glucose monitoring algorithm to prevent diabetes
complications. Applied Sciences, 10(3), 921.
https://doi.org/10.3390/app10030921
Vermesan, O., & Friess, P. (2013). Internet of things: Converging technologies
for smart environments and integrated ecosystems. River Publishers.
Vermesan, O., & Friess, P. (2014). Internet of things applications - from research
and innovation to market deployment. River Publishers.
VijiyaKumar, K., Lavanya, B., Nirmala, I., & Caroline, S. S. (2019). Random
Forest algorithm for the prediction of diabetes. 2019 IEEE International
Conference on System, Computation, Automation and Networking
(ICSCAN). https://doi.org/10.1109/icscan.2019.8878802
Vishnu, S., Ramson, S. R. J., & Jegan, R. (2020). Internet of medical things
(IOMT) - an overview. 2020 5th International Conference on Devices,
Circuits and Systems (ICDCS).
https://doi.org/10.1109/icdcs48716.2020.243558
Wadibhasme, P., Amin, A., Choudhary, P., & Saindane, P. (2020). Saathi—a
smart IOT-based pill reminder for IVF patients. Information and
Communication Technology for Intelligent Systems, 697–705.
https://doi.org/10.1007/978-981-15-7062-9_70
Walters, T. E., Nisbet, A., Morris, G. M., Tan, G., Mearns, M., Teo, E., Lewis, N.,
Ng, A. V., Gould, P., Lee, G., Joseph, S., Morton, J. B., Zentner, D.,
99
Sanders, P., Kistler, P. M., & Kalman, J. M. (2016). Progression of atrial
remodeling in patients with high-burden atrial fibrillation: Implications for
early ablative intervention. Heart Rhythm, 13(2), 331–339.
https://doi.org/10.1016/j.hrthm.2015.10.028
Wan, J., Al-awlaqi, M. A. A. H., Li, M. S., O’Grady, M., Gu, X., Wang, J., & Cao,
N. (2018). Wearable IOT enabled real-time Health Monitoring System.
EURASIP Journal on Wireless Communications and Networking, 2018(1).
https://doi.org/10.1186/s13638-018-1308-x
Wang, W., Li, J., Wang, L., & Zhao, W. (2012). The internet of things for resident
health information service platform research. IET International Conference
on Communication Technology and Application (ICCTA 2011).
https://doi.org/10.1049/cp.2011.0745
Westenberg, J. (2021, August 23). Samsung Galaxy Watch 4 review: A watch
built on promises. Android Authority. Retrieved October 25, 2021, from
https://www.androidauthority.com/samsung-galaxy-watch-4-review-
2731755/.
Wu, J. (2019, May 27). Building a collaborative filtering model with decision trees.
Medium. Retrieved November 5, 2021, from
https://medium.com/@jwu2/building-a-collaborative-filtering-model-with-
decision-trees-56256b95cb03.
Wu, T., Redouté, J. M., & Yuce, M. (2019). A Wearable, Low-Power, Real-Time
ECG Monitor for Smart T-shirt and IoT Healthcare Applications. In
100
Advances in body area networks: Post-conference proceedings of
bodynets (pp. 165–173). essay, SPRINGER.
Xie, J., Chen, Q., Shen, H., & Li, G. (2020). Review—wearable graphene devices
for sensing. Journal of The Electrochemical Society, 167(3), 037541.
https://doi.org/10.1149/1945-7111/ab67a4
Xin, Q., & Wu, J. (2017). A novel wearable device for continuous, non-invasion
blood pressure measurement. Computational Biology and Chemistry, 69,
134–137. https://doi.org/10.1016/j.compbiolchem.2017.04.011
Yaswanth, R., & Riyazuddin, Y. M. (2020). Heart disease prediction using
Machine Learning Techniques. International Journal of Innovative
Technology and Exploring Engineering, 9(5), 1456–1460.
https://doi.org/10.35940/ijitee.e2862.039520
YIN, Y., Zeng, Y., Chen, X., & Fan, Y. (2016). The internet of things in
Healthcare: An overview. Journal of Industrial Information Integration, 1,
3–13. https://doi.org/10.1016/j.jii.2016.03.004
Yue, X., Wang, H., Jin, D., Li, M., & Jiang, W. (2016). Healthcare data gateways:
Found Healthcare Intelligence on Blockchain with novel privacy risk
control. Journal of Medical Systems, 40(10).
https://doi.org/10.1007/s10916-016-0574-6
Zakaria, N. A., Mohd Saleh, F. N. B., & Razak, M. A. A. (2018). IOT (internet of
things) based infant body temperature monitoring. 2018 2nd International
101
Conference on BioSignal Analysis, Processing and Systems (ICBAPS).
https://doi.org/10.1109/icbaps.2018.8527408
Zhang, Q., Lu, J., & Jin, Y. (2020). Artificial Intelligence in Recommender
Systems. Complex & Intelligent Systems, 7(1), 439–457.
https://doi.org/10.1007/s40747-020-00212-w