Content uploaded by Muhammad Attique Khan
Author content
All content in this area was uploaded by Muhammad Attique Khan on Nov 10, 2022
Content may be subject to copyright.
Citation: Mirza, O.M.; Mujlid, H.;
Manoharan, H.; Selvarajan, S.;
Srivastava, G.; Khan, M.A.
Mathematical Framework for
Wearable Devices in the Internet of
Things Using Deep Learning.
Diagnostics 2022,12, 2750. https://
doi.org/10.3390/diagnostics12112750
Academic Editors: Leandro Donisi,
Carlo Ricciardi, Alfonso
Maria Ponsiglione,
Giuseppe Cesarelli, Paolo Gargiulo,
Milos Petrovic and Edda Maria
Capodaglio
Received: 23 September 2022
Accepted: 4 November 2022
Published: 10 November 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
diagnostics
Article
Mathematical Framework for Wearable Devices in the Internet
of Things Using Deep Learning
Olfat M. Mirza 1, Hana Mujlid 2, Hariprasath Manoharan 3, Shitharth Selvarajan 4,
Gautam Srivastava 5,6,7 and Muhammad Attique Khan 8, *
1
Department of Computer Science, College of Computers and Information Systems, Umm Al-Qura University,
Makkah 24381, Saudi Arabia
2Department of Computer Engineering, Faculty of Computer Engineering, Taif University,
Taif 21974, Saudi Arabia
3Department of Electronics and Communication Engineering, Panimalar Engineering College, Poonamallee,
Chennai 600123, Tamil Nadu, India
4Department of Computer Science, Kebri Dehar University, Kebri Dehar 001, Ethiopia
5Department of Mathematics and Computer Science, Brandon University, Brandon, MB R7A 6A9, Canada
6Research Center for Interneural Computing, China Medical University, Taichung 406040, Taiwan
7Department of Computer Science and Math, Lebanese American University, Beirut 1102, Lebanon
8Department of Computer Science, HITEC University, Taxila 47080, Pakistan
*Correspondence: attique.khan@hitecuni.edu.pk
Abstract:
To avoid dire situations, the medical sector must develop various methods for quickly and
accurately identifying infections in remote regions. The primary goal of the proposed work is to
create a wearable device that uses the Internet of Things (IoT) to carry out several monitoring tasks.
To decrease the amount of communication loss as well as the amount of time required to wait before
detection and improve detection quality, the designed wearable device is also operated with a multi-
objective framework. Additionally, a design method for wearable IoT devices is established, utilizing
distinct mathematical approaches to solve these objectives. As a result, the monitored parametric
values are saved in a different IoT application platform. Since the proposed study focuses on a multi-
objective framework, state design and deep learning (DL) optimization techniques are combined,
reducing the complexity of detection in wearable technology. Wearable devices with IoT processes
have even been included in current methods. However, a solution cannot be duplicated using
mathematical approaches and optimization strategies. Therefore, developed wearable gadgets can be
applied to real-time medical applications for fast remote monitoring of an individual. Additionally,
the proposed technique is tested in real-time, and an IoT simulation tool is utilized to track the
compared experimental results under five different situations. In all of the case studies that were
examined, the planned method performs better than the current state-of-the-art methods.
Keywords: wearable devices; Internet of Things (IoT); deep learning (DL); medical applications
1. Introduction
The advancements in medical applications that are increasing for day-to-day life are
greatly needed for monitoring groups of individuals at remote locations. Since most users
are using updated network information, designing a wireless device embedded within
apparel is much simpler. Whenever a monitoring device is designed with the proper
apparel, all individuals can carry it to any location, and it is possible to monitor their health
conditions at remote locations. Since remote management and monitoring are managed,
it is necessary to accumulate all monitored parameters on a cloud platform. Hence, an
Internet of Things (IoT) procedure is carried out where different types of infections are
examined, and all threshold values are stored on the cloud. Most wearable devices can
be incorporated without difficulty in all wireless communication devices, but, in turn, the
effects produced by the device for all individuals will increase.
Diagnostics 2022,12, 2750. https://doi.org/10.3390/diagnostics12112750 https://www.mdpi.com/journal/diagnostics
Diagnostics 2022,12, 2750 2 of 17
Early diagnosis of any infection needs to be monitored by observing all significant
indications that are present in the entire body’s functionality. Whenever wearable devices
are designed, as time progresses a higher computational load can be required, as they
are related to individual characteristics, whereas, in current-generation wireless networks,
microsensor-based chips are used, lacking computational power. As a result of microchips,
the efficiency of monitoring devices will increase, and the cost of implementation will
be reduced. However, the monitoring features of microsized sensor chips are nearly the
same as those of devices with large-scale deployment, though the loss and error functions
are reduced.
To transform the procedure of monitoring the state conditions and to reduce the
errors in identification processes, the proposed method introduces a new state condition
that reduces the monitoring time for all types of infections. After monitoring the exact
infections using an image processing technique, additional storage features are provided
in the proposed method as can be seen in the block diagram of integration shown in
Figure 1. From Figure 1, it is observed that wireless devices are managed with modernized
technologies. Thus, it is possible to establish a remote monitoring system. Once the device
parameters are set, the monitoring location is processed at frequent intervals using the
introduced dataset, which contains device and sensor data. Then, the data is pre-processed
using image identification parameters and data training is completed at this phase. During
the training phase, all devices are labelled, providing a complete interface without errors.
Hence, the complete wearable device model is represented with a low error rate, and loss
representations are greatly reduced in the system. Finally, data is stored in the cloud, and
output units are integrated into the observation process.
Diagnostics 2022, 12, x FOR PEER REVIEW 2 of 19
examined, and all threshold values are stored on the cloud. Most wearable devices can be
incorporated without difficulty in all wireless communication devices, but, in turn, the
effects produced by the device for all individuals will increase.
Early diagnosis of any infection needs to be monitored by observing all significant
indications that are present in the entire body’s functionality. Whenever wearable devices
are designed, as time progresses a higher computational load can be required, as they are
related to individual characteristics, whereas, in current-generation wireless networks,
microsensor-based chips are used, lacking computational power. As a result of micro-
chips, the efficiency of monitoring devices will increase, and the cost of implementation
will be reduced. However, the monitoring features of microsized sensor chips are nearly
the same as those of devices with large-scale deployment, though the loss and error func-
tions are reduced.
To transform the procedure of monitoring the state conditions and to reduce the er-
rors in identification processes, the proposed method introduces a new state condition
that reduces the monitoring time for all types of infections. After monitoring the exact
infections using an image processing technique, additional storage features are provided
in the proposed method as can be seen in the block diagram of integration shown in Figure
1. From Figure 1, it is observed that wireless devices are managed with modernized tech-
nologies. Thus, it is possible to establish a remote monitoring system. Once the device
parameters are set, the monitoring location is processed at frequent intervals using the
introduced dataset, which contains device and sensor data. Then, the data is pre-pro-
cessed using image identification parameters and data training is completed at this phase.
During the training phase, all devices are labelled, providing a complete interface without
errors. Hence, the complete wearable device model is represented with a low error rate,
and loss representations are greatly reduced in the system. Finally, data is stored in the
cloud, and output units are integrated into the observation process.
Figure 1. The architecture of the proposed wearable IoT device.
1.1. Literature Survey
In this section, all pertinent existing models are examined to identify their shortcom-
ings and any potential improvements that could be made to the wearable IoT device by
altering a few parametric monitoring systems. Most current models incorporate the same
dataset during the observation stage, yet the evaluation results for created systems can be
very diverse. Thus, a standard representation form is required, which is given after the
survey model. Additionally, as medical applications require a fundamental
Figure 1. The architecture of the proposed wearable IoT device.
1.1. Literature Survey
In this section, all pertinent existing models are examined to identify their shortcom-
ings and any potential improvements that could be made to the wearable IoT device by
altering a few parametric monitoring systems. Most current models incorporate the same
dataset during the observation stage, yet the evaluation results for created systems can be
very diverse. Thus, a standard representation form is required, which is given after the
survey model. Additionally, as medical applications require a fundamental understanding
of all issues, recent research is also reviewed. More details about next-generation networks
and their real-time implementation method, which is entirely dependent on network con-
ditions, are given in [
1
]. According to the implementation plan, to increase the general
Diagnostics 2022,12, 2750 3 of 17
acceptance rate, it is crucial to assess the growth of products before exposing them to the
market. However, one of the disadvantages of next-generation networks is that wear-
able devices must be synced with all network-related criteria, which is very challenging.
Therefore, several applications related to developing body sensor networks that offer basic
designs are provided to tackle such synchronization scenarios [
2
]. Numerous technical
specifications are given concerning antenna design in the included designed model, but
no separation cases are assessed for various applications. When fog layers are utilized
in the identification phases, wearable IoT is incorporated even for detecting the presence
of hazardous diseases at an early stage [
3
]. Fog layers serve as a middleman between
all agencies and the cloud, accumulating storage space for individual reports. However,
because fog layers demand high bandwidth for transmission and reception, they can also
be employed in other identification and diagnosis processes.
Wearable devices must be manufactured following the intended application platform
if they are to be implanted in human bodies [
4
]. To detect the performance of different
users, a separate record must be added to the fabrication procedures; as a result, a data
record is added to the system. Even though the working model has been fully sensitized,
a significant flaw is that not all applications have access to the fabrication conditions in
a proper manner. An offloading strategy that boosts the energy of node systems has
been developed to address some of the difficulties in fabrication processes [
5
]. Smart
wearable gadgets have a greater capacity for storing significant data amounts and have
strong offloading techniques. A distinct platform is generated due to the data storage
method, raising the system’s overall operating cost. A wearable gadget was created to
detect Alzheimer’s disease in [
6
]. A large amount of individual storage is needed because
the detection method typically involves processing more data. Separate storage spaces
cause various applications to be segregated, which makes it much more challenging to
process in real-time. Some of the enabling technologies for the Internet of Medical Things
(IoMT) have been developed, and serve the entire environment by identifying diseased
individuals within specific distance limits [
7
]. Distance restrictions make delivering a
higher data rate for a single data transfer at the transmitter end challenging. A standard
technique is required to monitor all required system parameters; even with an increase in
data flow, a standard approach is needed.
Even if wireless IoT devices can monitor people’s health, it is still essential to examine
the affected person’s allocated space for diagnosis in [
8
]. An original mathematical model is
constructed to test the space allocation, allowing for the distribution of personal space even
during times of high demand. Formulations of this kind are used to make critical judgments
during times of high risk. The data from wearable devices is also checked using a two-stage
paradigm, where accelerometer sensors track all physical activity in [
9
]. Since minute-to-
minute readings are measured as part of this monitoring approach, both the time-varying
and invariant parameters are tracked. Even so, both variables produce data with a high
density, so the counting procedure is crucial to the processing stage. A posture prediction
procedure is carried out using a fusion-based model to avoid high-density data, resulting in
error-free data transfer and a significant improvement in posture-prediction accuracy [
10
].
However, there are a variety of postures that need to be trained using storage techniques.
Hence, the training and testing phases call for a collection method involving gathering
sensor data. By enhancing the security of each data collection conducted in various physical
situations, wearable IoT data transmission accuracy may be increased, as shown in [
11
].
An automatic monitoring system employing Artificial Intelligence (AI) is introduced as
environmental factors change, encouraging cyber–physical systems’ development. Even
where a step count strategy is used, it is possible to improve the quality of service for all
end users of all wireless IoT devices, by using a defined mathematical approach [12].
A fog model is again required to be incorporated in several situations, but, in addition
to the procedures above, commonly specified methodologies are represented [
13
–
16
] using
an architecture-described system. In the next section, we will discuss a wearable IoT
Diagnostics 2022,12, 2750 4 of 17
model that uses analytical equations to determine what is wrong with a person within
a certain period.
1.2. Research Gap and Motivation
All the existing methods, as indicated in Table 1, focus on any one major objective
by using medical healthcare services as one of the application platforms. However, if
existing methods use wearable devices, then some of the parameters that are needed for
time-based monitoring need to be provided. Even existing models need to provide a precise
mathematical approach that makes the medical healthcare system function more accurately.
However, some of the introduced mathematical methods provide information related to
cloud-based services, where the monitored data from medical systems are transmitted,
with security features. Further, if a user needs to identify a particular infection at a
remote location, additional parameters are required, which is a significant gap in the
existing models. Whenever medical healthcare systems are integrated with technology
developments using wireless networks, the monitored characteristic features must provide
accurate outcomes, and the storage system must function appropriately.
Table 1. Background and objectives (existing vs. proposed).
References Background Objectives
[16]Overview of medical devices and application
platform for wearable devices Minimize the cost of implementation
[17]Detection of interruptions that are present in
monitoring systems for wireless devices Minimize the number of IoT data interruption
[18]Possible developments in wireless network
applications for medical healthcare Minimization of cost
[19]Fabrication design of wearable devices for different
applications Minimization of congestion
[20]Design of wideband antennas for wireless
communication transfer Maximization of coverage
Proposed Deep learning approach for wearable devices Multi-objective framework with minimization of
loss, energy and errors
To overcome the gap present in existing models, our proposed system is introduced,
where all monitoring is processed only under certain device constraints. A mathematical
model is formulated for a device-monitoring system in medical healthcare applications.
Even the procedure of device testing is made only by using loop formation provided by
a designed mathematical approach. In addition, the proposed method is integrated with
a deep learning (DL) algorithm; thus, efficiency at the output unit is augmented with a
maximized monitoring distance. Since devices operate at a considerable distance, most of
the characteristic features of infections are observed without installing wearable devices
in the frame measures. Furthermore, all errors during the transmission process will be
minimized; thus, the device functions more accurately.
1.3. Contributions
The technique of a wireless monitoring system for healthcare applications, which is
designed in the proposed method with wearable devices, is used for solving three major
objectives as follows:
•
Maximize the monitoring device’s distance, so all infections are identified without
connecting to the frame;
•
Minimize all errors present in the infection identification and data transmission process,
thus increasing the efficiency of the device;
•
Integrate a deep learning model for reducing the loss of designed devices with a
unique representation of mathematical models.
Diagnostics 2022,12, 2750 5 of 17
1.4. Paper Organization
The rest of the paper is organized as follows: Section 2provides a formulation of
a system model with the appropriate variables. Section 3integrates the optimization
framework with step-by-step implementation. Section 4examines the combined outcomes
of the proposed method with system formulations, and comparison case studies are also
analyzed. Finally, Section 5concludes with the advantages of the proposed method for
future work.
2. System Model
This portion, where the designed model is directly incorporated into a real-time
setup, formulates the analytical model of wearable IoT devices utilized for recognizing
various viruses and infections. Since the classification technique is used to identify people’s
ailments, it is crucial to offer resources connected to multiple materials. As a result, the
identification procedure is split into two distinct units, each made up of identifiable and
unidentified users, where the infirmary centres receive central data. Additionally, wearable
IoT devices classify the types of viruses based on blood samples and send the information
to local emergency rooms. So, to figure out how far the emergency rooms and hospitals are
from each other [5], we can use Equation (1).
disti=max
n
∑
i=1
Ai+Pi+ti+hi
ρi
(1)
In Equation (1), the objective function where the identification distance is maximized.
Equation (2) can be used to define the energy representation of local computation tasks
when wearable devices with unloading tasks are used [5,21].
Ei=min
n
∑
i=1
(E1+. . . +Ei)∗δi(2)
Equation (2) is a reduction function that reduces the workload of each device, because
numerous devices are utilized to identify various illnesses, necessitating a greater number
of wearables. However, since the device’s data functions are connected by a wireless
module, Equation (3) is used to measure the following data waiting time [7,22].
Waiti=min
n
∑
i=1
tph ∗Ecm (3)
To illustrate how connected devices work, the third objective function, which is stated
as a minimization function, is employed. Equation (4) is used to represent the energy of the
communication module [6].
Ecm =
n
∑
i=1
pcm
βi
∗dn(i)(4)
Equation (4) only applies to situations where people are infected by dangerous viruses;
in all other situations, the energy is fully shut off. Equation (5) is used to calculate the
sensitivity factor under various ecological conditions [9].
Si=min
n
∑
i=1
probiHi
eventsi(5)
In Equation (5), the number of occurrences shows that control measures using wearable
IoT devices, which are built using Equation (6), must be put in place if a person has
been infected [12].
RIi=
n
∑
i=1
Ir(i)
Tr(i)∗100 (6)
Diagnostics 2022,12, 2750 6 of 17
Equation (7) is used for wearable devices to figure out the weighted combination of
variables or quality factors since different parameters are used in the representation process [
5
].
Qf(i) = max
n
∑
i=1
γi∗Dq(i)(7)
All of the stated Equations (1)–(7), are used to represent the effects of wearable tech-
nology and have an optimization mechanism built in to increase operating efficiency. The
variables used in the equations are summarized in Table 2. This optimization process is
covered in more detail in the next section.
Table 2. Mathematical notations.
Variables Description
AiIdentified area of infection
PiDifferent types of patients with infections
ti,hiTime period and infirmaries of identification
ρiDemand for persistence in a particular area
E1+. . . +EiEnergies of different wearable devices
δiWork functionality of devices
tph Time period of mobile-connecting devices
Ecm Energy of communication module
pcm Power delivered to the communication module
βiThroughput of the device module
dnSize of data to be transmitted
HiGroup of affected users
eventsiOccurrence of different events
Ir,TrIndividuals in infected regions and total covered regions
γiRelative weights of all parameters
DqData quality index
probg,probdProbability of generator and data
ls,lpSample loss and prediction loss periods
ToObject representation target
S1..Si. . . SnWearable signal-representation matrix
3. Optimization Algorithm
It is crucial to introduce an optimization method with efficient mathematical mea-
surements to discover the risk analysis of examined medical photographs. As a result,
the described system model is merged with DL which produces beneficial effects by dis-
criminator combinations [
23
,
24
]. The main benefit of DL in medical imaging is that it
significantly lowers the risk of identification since it makes it easier to identify more infec-
tious diseases using a set of implicit functions. Additionally, it is crucial to create a feature
automation procedure for diagnosis, so that a stated logical structure can be implemented
without causing any side effects. Additionally, generative measurements have a vastly
improved complexity of conjunction for supplying matching features compared to standard
measurement types. Therefore, even for big dataset models where identification time is
minimized, a fixed distribution set is used in the identification processes. However, as
shown in Equation (1), the data distribution model utilizing DL can be controlled using
training samples, as indicated in Equation (8) [21].
trains(i) = min
n
∑
i=1
probg(i) + probd(i)(8)
Equation (8) shows that the probability function of DL is a minimization problem,
where the wearable object spreads images it has collected. Since it is possible to find
Diagnostics 2022,12, 2750 7 of 17
loss functions during these transmissions, it is essential to use Equation (9) to minimize
data loss [22].
lossi=min
n
∑
i=1
ls(i) + lp(i)(9)
To reconstruct the target system using several wearable items, the loss periods must be kept
to a minimum. So, Equation (10) is used to measure the signal matrix of wearable devices [
21
,
22
].
Oi=
n
∑
i=1
To(i) + S1S2
SiSn(10)
Equation (10) states that the signal representation matrix for loss functions must be
reduced to a minimum of 0.01%, making it possible for wearable devices to communicate
without interruption, even at greater separation distances. Figure 2shows the integration
method’s careful, step-by-step application utilizing DL.
Step-by-Step Implementation of DL Using Adversarial Networks
Input: Initialize the maximum distance in terms of identified area, type of patients,
periods, and demand for capturing images in wearable devices using the representative
values of the matrix functions Ai(Ai≤i≤n),Pi(P≤i≤n)and the time matrix ti;
Output: Minimize the load and waiting time and maximize the distance and quality of
service factors:
Step 1: The objective function is constructed with different types of patients’ data and
one demand persistent value using ρi;
Step 2: Establish the relationship between the different energy representations and
working functionalities of devices that must be followed by waiting periods
Waiti
with
1
≤i≤n
, and its wireless module representation values
Ecm
at different marginal functions;
Step 3: while (disti<N)do.
•
Select the captured images and measure the communication module energy representa-
tion values using different data sets in a systematic way for computing the throughput,
by using Equation (3);
•Verify the value of Siand Ecm using the probability value set;
•If the complexities of identification are higher, Siis not at (Si<N)do;
•
Divide the probability of different events using the number of affected users
Hi
and
eventsi
, which ensures different ecological conditions using Equation (5)
Hi
with
1≤i≤Ninto Nnumber of affected points; //Training sample phase
•
Update the infected region and total region areas using a spreading ratio matrix with
minimized loss function using sample and prediction loss as shown in Equation (9);
//Loss phase
•
Select the number of loss functions with wearable signal representation values of
different training samples with separate image analyses in a single output
trains(i)
, as
defined in Equation (8);
•
Update the object representation values of the relative positions and identified position,
followed by measurement of the data quality index, and compute the relative weight
of all parameters γi, as defined in Equation (7);
•
Identified infections at each point are updated by using the probability of the generator
and data that is designed for the image signals;
Ni(new)=Ni(old)+1;
End;
Step 4: If
(lossi<Ni)
then
lossi←0
; //Interchange the existing solution in the
current loop with the new solution; End if;
Step 5: If
(disti[0, 1]<lossi)
then Re-initialize the images that are taken with the new
processing technique; Obtain the overall best solution; End if;
Diagnostics 2022,12, 2750 8 of 17
Step 6: If (Qf(i)>N)//Existing solution is replaced with the new solution
Nnew =Ni;
Nold =Ni;
//Attain the most feasible solutions for determining the overall best solution; Incre-
ment the count Niby 1; Return the best overall solution; End.
Diagnostics 2022, 12, x FOR PEER REVIEW 8 of 19
Figure 2. Flowchart of DL for wearable healthcare applications.
Figure 2. Flowchart of DL for wearable healthcare applications.
Diagnostics 2022,12, 2750 9 of 17
4. Experimental Outcomes
For wearable IoT devices, the integrated process of a defined system model and
optimization algorithm must be tested in real-time using carefully defined experimental
models. To identify various infections, real-time verification with a simulation setup is
used, and the parametric values that help enhance the effectiveness of the suggested system
are offered. All conditions are detected in all developed device models, which is the main
reason for merely supplying an experimental value. However, it is crucial to identify which
wearable device performs more effectively than the other devices in various parametric
values. Additionally, the entire dataset is provided in the input layer of the suggested
methodology. This is only gathered from multiple treatment facilities because it is much
more accurate than other data set representations. Further, a real-time application is created,
where all defects in the application segments are fixed after extensive testing and simulation
over more than 1000 phases.
Additionally, the proposed technique incorporates extra data based on the availability
of beds at all surrounding hospitals during times of emergency. As a result, if any infections
are seriously impacted, the wearable device will be able to detect them and, using location
data, alert local treatment facilities. In real-time, the visualization process of wearable
devices is carried out by examining the analytical framework under five different scenarios,
as follows:
Scenario 1: Measurement of distance;
Scenario 2: Waiting periods;
Scenario 3: Sensitivity factors;
Scenario 4: Quality of measurement;
Scenario 5: Loss periods.
All the input parameters used for designing a particular wearable device are provided
in Table 3. The primary application of the proposed wireless wearable device, which
operates with a DL algorithm, is to observe the type of infection imposed on an individual.
In current changing environments, most people must undergo regular medical checkups
to safeguard themselves from different types of infected viruses in the system. However,
due to the lack of remote monitoring systems, it is possible to continuously monitor all the
infections quickly, so that the lives of surrounding people can also be saved. Hence, the
proposed method is applied in medical applications to discover all infected viruses in an
anthropological frame. All the values represented in the simulation tables are achieved
only after careful experimentation and testing of wearable devices. In addition, all defined
values denote the device’s performance measures that operate under proper environmental
conditions. The representation values are simulated using the MATLAB IoT toolbox for
better empathetic constraints. Thus, all parameters, including the sensitivity of the designed
system, are analyzed. However, to provide a real-time representation comparison, the
form of the figures is also simulated in MATLAB using a 3-dimensional (3D) plot. The
simulated plots indicate that the proposed method performs much better than existing
techniques regarding various parameters, where all affected figures are described using
loop-based codes. For simulating corresponding representations, a hardware processing
unit is directly connected to MATLAB using a serial cable port. Therefore, the plot is
simulated in MATLAB with a DL integration process.
All primary scenarios are integrated into IoT-based hardware and software platforms
using an MIT application intervention. Still, all output units are directly connected with the
MATLAB IoT processing tool, as simulation characteristics with a comparison state need to
be analyzed. In the application platform, supporting blocks for wearable IoT devices are
created and connected with the user interface; thus, the application layer’s functionalities
are introduced. During this type of functionality implementation, all properties of wear-
able devices are added at the initial phase of application creation using proper list-view
selections. The outcomes are presented for all observed periods since all parametric values
cannot be measured if findings are evaluated in subsequent weeks. Additionally, the study
Diagnostics 2022,12, 2750 10 of 17
results offer potential low-cost ways to improve the physiological health of all patients.
The following are comprehensive descriptions of each scenario.
Table 3. Input specifications (existing vs. proposed).
Key Features Existing [3] Proposed
Package size 4 ×5.2 ×1.3 2 ×2×0.7
Sensor power High power greater than 5 volts
Ultra-low power with a three-axis accelerometer
Noise density 50 22
Current consumption 0.89 mA 0.55 mA
Maximum distance 2.5 m 5.7 m
Sensitivity 15 2
Gain bandwidth 4 kHz 8 kHz
Memory unit 100 GHz 500 GHz
Run mode 12 microamperes 30 microamperes
4.1. Scenario 1
Wearable-device distance measurement is crucial because some people will keep their
devices close to their bodies if they are particularly sensitive to certain membrane regions.
Therefore, it is crucial to gauge the distance at which all users are permitted to remove
their wearable technology and store it. The proposed method clarifies distance by looking
at the afflicted areas, where certain illnesses can be detected in specific locations. In the
observed technique, identification with information measurement time is also considered
a summation value. Another unique feature of the suggested method is that total values
are separated by measuring the distance of demand in a specific area. Furthermore, fitted
wearable devices are not physically harmed while measuring distance, conforming to a
standard pattern for distance measurements. After separating all new values by demand,
Figure 3shows a simulation of distance measurement.
Diagnostics 2022, 12, x FOR PEER REVIEW 12 of 19
Figure 3. Distance and demand measurements.
Table 4. Raised demand and distance.
Number of Infected Areas
Demand
Distance [3]
Distance (Proposed)
120
32
1.3
1.6
180
44
1.7
2.8
260
57
2
3.4
340
61
2.2
4.9
400
65
2.5
5.7
4.2. Scenario 2
Wearable technology that transmits data must offer a low energy value for sending
a single data packet to a destination. As a result, the rate of energy measurement is seen
for various wearable devices, where it must be reduced while maintaining functionality
in specified devices. The waiting time for all data will be minimized, using the minimiza-
tion objective function. A wireless communication module was introduced for this sort of
reduction target, significantly reducing data transfer time. The communication module
can also be created using a low-cost design that considers the lifespan of all required com-
ponents. As a result, the wireless module’s reproduction rate and period will give precise
results linked to the waiting period for the data that are transmitted by each user end. The
simulation results of waiting times for various wearable devices are shown in Figure 4.
Figure 3. Distance and demand measurements.
Diagnostics 2022,12, 2750 11 of 17
Table 4. Raised demand and distance.
Number of Infected Areas Demand Distance [3] Distance (Proposed)
120 32 1.3 1.6
180 44 1.7 2.8
260 57 2 3.4
340 61 2.2 4.9
400 65 2.5 5.7
4.2. Scenario 2
Wearable technology that transmits data must offer a low energy value for send-
ing a single data packet to a destination. As a result, the rate of energy measurement
is seen for various wearable devices, where it must be reduced while maintaining func-
tionality in specified devices. The waiting time for all data will be minimized, using the
minimization objective function. A wireless communication module was introduced for
this sort of reduction target, significantly reducing data transfer time. The communica-
tion module can also be created using a low-cost design that considers the lifespan of all
required components. As a result, the wireless module’s reproduction rate and period
will give precise results linked to the waiting period for the data that are transmitted by
each user end. The simulation results of waiting times for various wearable devices are
shown in Figure 4.
Diagnostics 2022, 12, x FOR PEER REVIEW 13 of 19
Figure 4. Waiting periods at high energy.
Figure 4 and Table 5 show that the periods of the devices vary from 10, 20, 30, 40, and
50 s, with the number of energy modules being represented for each period in steps of
100. The devices’ waiting times are calculated after the reproduction rate is measured, to
ensure the least amount of waiting time possible. In the comparative state, the existing
approach from [3] offers a single data segment with a higher waiting time than the sug-
gested way. This may be tested for 30 s using 500 energy modules, where the total time
spent in the unload condition for the proposed technique is equivalent to 1.1 s. However,
with the identical setup, the current approach only delivers 2.17 s, so an increase in wait-
ing time is noted. Inadequate data proportions exist in the present method due to in-
creased waiting time. However, this problem is resolved in the proposed method by em-
ploying adversarial tactics.
Table 5. Energy modules with waiting periods.
Period
Number of Energy Mod-
ules
Waiting Period [3]
Waiting Period (Pro-
posed)
10
100
2.33
1.25
20
300
2.21
1.16
30
500
2.17
1.1
40
700
2.06
1
50
900
2
0.8
4.3. Scenario 3
In this case, probability analysis estimates the sensitivity parameter by counting the
number of incidents. The number of infected people is directly separated by utilizing
event-measurement cases for sensitivity measurements in wearable IoT devices. Reduced
device sensitivity is crucial as a safety precaution, because wearable technology may harm
all users if the sensitivity factor is higher. Additionally, since direct radiation will be pre-
sent across the entire network, it is necessary to eliminate all direct radiation from the
system-generating process so that sensors subject to highly sensitive conditions will re-
main intact. Additionally, since the probability of measurement alterations is significantly
larger when an entire network is infected, the system must be modified to eliminate
Figure 4. Waiting periods at high energy.
Figure 4and Table 5show that the periods of the devices vary from 10, 20, 30, 40,
and 50 s, with the number of energy modules being represented for each period in steps
of 100. The devices’ waiting times are calculated after the reproduction rate is measured,
to ensure the least amount of waiting time possible. In the comparative state, the exist-
ing approach from [
3
] offers a single data segment with a higher waiting time than the
suggested way. This may be tested for 30 s using 500 energy modules, where the total
time spent in the unload condition for the proposed technique is equivalent to 1.1 s. How-
ever, with the identical setup, the current approach only delivers 2.17 s, so an increase in
waiting time is noted. Inadequate data proportions exist in the present method due to
increased waiting time. However, this problem is resolved in the proposed method by
employing adversarial tactics.
Diagnostics 2022,12, 2750 12 of 17
Table 5. Energy modules with waiting periods.
Period Number of Energy Modules Waiting Period [3] Waiting Period (Proposed)
10 100 2.33 1.25
20 300 2.21 1.16
30 500 2.17 1.1
40 700 2.06 1
50 900 2 0.8
4.3. Scenario 3
In this case, probability analysis estimates the sensitivity parameter by counting the
number of incidents. The number of infected people is directly separated by utilizing
event-measurement cases for sensitivity measurements in wearable IoT devices. Reduced
device sensitivity is crucial as a safety precaution, because wearable technology may harm
all users if the sensitivity factor is higher. Additionally, since direct radiation will be
present across the entire network, it is necessary to eliminate all direct radiation from
the system-generating process so that sensors subject to highly sensitive conditions will
remain intact. Additionally, since the probability of measurement alterations is significantly
larger when an entire network is infected, the system must be modified to eliminate
mobile devices’ sensing capabilities. Even though the probability analysis model uses
separation techniques, it is still possible to regulate the full spread ratio by looking at all
the infected areas. Figure 5shows the comparison scenario and simulation results for
the sensitivity parameters.
Diagnostics 2022, 12, x FOR PEER REVIEW 14 of 19
mobile devices’ sensing capabilities. Even though the probability analysis model uses sep-
aration techniques, it is still possible to regulate the full spread ratio by looking at all the
infected areas. Figure 5 shows the comparison scenario and simulation results for the sen-
sitivity parameters.
Figure 5. Measurements of sensitivity.
Figure 5 and Table 6 make it clear that the number of impacted users is far higher, so
only a small number of affected cases—1000, 2000, 3000, and 5000—are taken into account.
For each affected user, the probability of occurrence is computed as 40, 27, 35.56, and 73,
respectively. Since both infected and total regions are observed for each probability of
occurrence, individual probability values are minimized. Furthermore, it makes sense in
a comparison situation that the existing method’s sensitivity percentage is substantially
higher than our proposed method’s. This may be demonstrated by using the 5000 affected
users and a high repeated probability of 73; in this scenario, the proposed method’s sen-
sitivity is just two percentage points, but the proposed method’s sensitivity is 15 percent-
age points. The cost of measurements goes up, since the existing process, which is less
sensitive than the suggested method, requires more control measures.
Table 6. Percentage of sensitivity.
Number of Af-
fected Users
Probability of
Occurrence
Percentage of Sen-
sitivity [3]
Percentage of Sensitivity (Pro-
posed)
1000
40
31
20
2000
27
26
14
3000
35
21
10
4000
56
17
5
5000
73
15
2
4.4. Scenario 4
Every developed communication equipment used in various applications must offer
high-quality service to every network user. Similar to how monitoring needs to be im-
proved, intended wearable IoT devices must provide high-quality service expressed in
terms of data. Therefore, a mathematical model is created to enhance the service quality
Figure 5. Measurements of sensitivity.
Figure 5and Table 6make it clear that the number of impacted users is far higher,
so only a small number of affected cases—1000, 2000, 3000, and 5000—are taken into
account. For each affected user, the probability of occurrence is computed as 40, 27,
35.56, and 73, respectively. Since both infected and total regions are observed for each
probability of occurrence, individual probability values are minimized. Furthermore, it
makes sense in a comparison situation that the existing method’s sensitivity percentage is
substantially higher than our proposed method’s. This may be demonstrated by using the
5000 affected users and a high repeated probability of 73; in this scenario, the proposed
method’s sensitivity is just two percentage points, but the proposed method’s sensitivity is
Diagnostics 2022,12, 2750 13 of 17
15 percentage points. The cost of measurements goes up, since the existing process, which
is less sensitive than the suggested method, requires more control measures.
Table 6. Percentage of sensitivity.
Number of Affected
Users
Probability of
Occurrence
Percentage of
Sensitivity [3]
Percentage of
Sensitivity (Proposed)
1000 40 31 20
2000 27 26 14
3000 35 21 10
4000 56 17 5
5000 73 15 2
4.4. Scenario 4
Every developed communication equipment used in various applications must offer
high-quality service to every network user. Similar to how monitoring needs to be im-
proved, intended wearable IoT devices must provide high-quality service expressed in
terms of data. Therefore, a mathematical model is created to enhance the service quality
by considering the relative importance of all the system factors. The relative weights from
Table 7are significantly different from the present procedure because, in earlier circum-
stances, only data weights were considered without any specified formulations. The total
reproduction rate improves the service because it measures the system’s relative weight
and the data quality index at the transmitter end.
Table 7. Quality of service with a relative weighting factor.
Relative Weights Percentage of Data
Quality Quality of Service [3]Quality of Service
(Proposed)
10 45 60 75
15 54 62 79
20 69 63 82
25 75 63 84
30 82 63 84
Even though it is possible to measure service quality using a variety of control mea-
sures, the data quality index for wearable devices will decrease if controls are implemented;
so, in the proposed method, controls are provided in the form of a relative weight factor.
The simulated result of service quality is shown in Figure 6. It is reasonable to assume that
the proposed method will provide a much higher service quality than the current approach,
based on Figure 5and Table 7[
3
]. To simulate the quality-of-service measurement, weights
of 10, 15, 20, 25, and 30 g are considered. These weights include microsensors made of
light materials. The percentage of data quality provided during such an operation is 45,
54, 69, 75, and 82, respectively, and the rates above are replicated in the quality-of-service
factor. When the reproduction rate is measured and compared to the existing method, the
latter yields a service rate of 84% for high relative weights and data quality, compared
to 63% for the exact requirement for consistent quality of service as the relative weights
rise. Low parametric determinations also cause the low quality of the index given by the
current method. However, the proposed method can improve the quality of service for
high measurement values.
Diagnostics 2022,12, 2750 14 of 17
Diagnostics 2022, 12, x FOR PEER REVIEW 15 of 19
by considering the relative importance of all the system factors. The relative weights from
Table 7 are significantly different from the present procedure because, in earlier circum-
stances, only data weights were considered without any specified formulations. The total
reproduction rate improves the service because it measures the system’s relative weight
and the data quality index at the transmitter end.
Even though it is possible to measure service quality using a variety of control
measures, the data quality index for wearable devices will decrease if controls are imple-
mented; so, in the proposed method, controls are provided in the form of a relative weight
factor. The simulated result of service quality is shown in Figure 6. It is reasonable to as-
sume that the proposed method will provide a much higher service quality than the cur-
rent approach, based on Figure 5 and Table 7 [3]. To simulate the quality-of-service meas-
urement, weights of 10, 15, 20, 25, and 30 g are considered. These weights include mi-
crosensors made of light materials. The percentage of data quality provided during such
an operation is 45, 54, 69, 75, and 82, respectively, and the rates above are replicated in the
quality-of-service factor. When the reproduction rate is measured and compared to the
existing method, the latter yields a service rate of 84% for high relative weights and data
quality, compared to 63% for the exact requirement for consistent quality of service as the
relative weights rise. Low parametric determinations also cause the low quality of the
index given by the current method. However, the proposed method can improve the qual-
ity of service for high measurement values.
Figure 6. Quality of service measurements.
Table 7. Quality of service with a relative weighting factor.
Relative
Weights
Percentage of Data
Quality
Quality of Service [3]
Quality of Service
(Proposed)
10
45
60
75
15
54
62
79
20
69
63
82
25
75
63
84
30
82
63
84
4.5. Scenario 5
Figure 6. Quality of service measurements.
4.5. Scenario 5
In this scenario, the loss of measurement in wearable IoT devices is measured and
handled as significant parametric real-time measurements. After the detection process and
measurements are taken to reduce the amount of duplicate data in the system, it is crucial to
understand the amount of data transmitted. All the detected data must be represented in the
output system to prevent the loss of the input data measurements since wearable devices are
used in medical diagnosis. However, real-time data loss will occur; as a result, adversarial
deep learning optimization is included in the proposed method to detect the sample and
prediction loss periods. However, training samples of the generator system must be measured
precisely, where inputs given in various arrangements will be processed in a standard mode
to detect such losses in the system. The proposed system’s data loss is significantly reduced
due to common mode representation, and its simulation results are shown in Figure 7.
Diagnostics 2022, 12, x FOR PEER REVIEW 16 of 19
In this scenario, the loss of measurement in wearable IoT devices is measured and
handled as significant parametric real-time measurements. After the detection process
and measurements are taken to reduce the amount of duplicate data in the system, it is
crucial to understand the amount of data transmitted. All the detected data must be rep-
resented in the output system to prevent the loss of the input data measurements since
wearable devices are used in medical diagnosis. However, real-time data loss will occur;
as a result, adversarial deep learning optimization is included in the proposed method to
detect the sample and prediction loss periods. However, training samples of the generator
system must be measured precisely, where inputs given in various arrangements will be
processed in a standard mode to detect such losses in the system. The proposed system’s
data loss is significantly reduced due to common mode representation, and its simulation
results are shown in Figure 7.
Figure 7. Comparison of loss factor.
Figure 7 and Table 8 show that there were 89, 124, 153, 205, and 279 data sample
losses for the input representations, respectively. The designed system predicts that the
following minimum losses will occur for each sample loss: 51, 87, 99, 125, and 154, respec-
tively. If the total loss is less than the expected loss, then the wireless IoT device performed
well in the data transmission cases. The total data loss of the proposed method is lower
when compared to the existing method [3], but good performance is only attained at high
sample data representations. This can be demonstrated by a sample loss of 279 and a pre-
dicted loss of 154, where the overall data loss is 142. Thus, it is possible to minimize data
loss and increase system throughput using deep adversarial learning. Even if a lot of data
are given, the total loss can still be less than expected [19,20].
Table 8. Loss measurement values.
Number of
Sample Loss
Number of Predicted Loss
Total Loss [3]
Total Loss (Pro-
posed)
89
51
117
76
124
87
203
87
153
99
245
103
205
125
305
128
279
154
400
142
Figure 7. Comparison of loss factor.
Diagnostics 2022,12, 2750 15 of 17
Figure 7and Table 8show that there were 89, 124, 153, 205, and 279 data sample losses
for the input representations, respectively. The designed system predicts that the following
minimum losses will occur for each sample loss: 51, 87, 99, 125, and 154, respectively. If
the total loss is less than the expected loss, then the wireless IoT device performed well
in the data transmission cases. The total data loss of the proposed method is lower when
compared to the existing method [
3
], but good performance is only attained at high sample
data representations. This can be demonstrated by a sample loss of 279 and a predicted
loss of 154, where the overall data loss is 142. Thus, it is possible to minimize data loss and
increase system throughput using deep adversarial learning. Even if a lot of data are given,
the total loss can still be less than expected [19,20].
Table 8. Loss measurement values.
Number of
Sample Loss
Number of
Predicted Loss Total Loss [3]Total Loss
(Proposed)
89 51 117 76
124 87 203 87
153 99 245 103
205 125 305 128
279 154 400 142
4.6. Performance Analysis
The performance analysis determines whether the designed devices function appro-
priately with unique characteristic features compared to other existing methods. Therefore,
in this section, the robustness characteristics of wearable devices are examined, and com-
parisons with existing models are also provided.
4.7. Robustness Characteristics
A wearable device that operates in IoT must be effective under all circumstances,
with low robust conditions. However, wearable devices are subject to 4% of robustness
if operated in a hostile environment, whereas in standard environmental cases, 1% of
robustness will be present [
4
]. The robustness of wearable devices with DL determines
the strength of identification when the data are transmitted in a particular system. If both
transmission and identification strengths are much higher, then high efficiency at output
units can be achieved. In addition, in wireless devices with communication media, the
most potent transmitters are not used; hence, the strength of identification is much lesser.
However, in the proposed method using deep learning, better optimization characteristics
are achieved. Therefore, the identification process is made much simpler. The robustness
characteristics of the proposed and existing method are provided in Figure 8.
From Figure 8, it can be observed that the strength of the proposed wearable devices
for identifying a particular infection is much higher compared to the existing method. To
verify the robustness characteristics, more iterations are considered from 10 to 100, where,
during each variation, the strength of identification changes between lower and upper
values. However, during these changes, the proposed method maintains tolerable limits, as
the threshold capacity of the device is predefined in the wearable device. However, in the
existing method [
3
], even if the low values are changed, tolerable limits are not maintained
in the system. As a result of low tolerable conditions, the robustness of the existing method
remained at 5%, while the proposed method provides 1% robustness to the maximum
extent. These comparison values are plotted concerning the simulation time standard in
MATLAB, but the proposed wearable devices provide much low robustness in real-time, as
the loss factor is minimized.
Diagnostics 2022,12, 2750 16 of 17
Diagnostics 2022, 12, x FOR PEER REVIEW 17 of 19
4.6. Performance Analysis
The performance analysis determines whether the designed devices function appro-
priately with unique characteristic features compared to other existing methods. There-
fore, in this section, the robustness characteristics of wearable devices are examined, and
comparisons with existing models are also provided.
4.7. Robustness Characteristics
A wearable device that operates in IoT must be effective under all circumstances,
with low robust conditions. However, wearable devices are subject to 4% of robustness if
operated in a hostile environment, whereas in standard environmental cases, 1% of ro-
bustness will be present [4]. The robustness of wearable devices with DL determines the
strength of identification when the data are transmitted in a particular system. If both
transmission and identification strengths are much higher, then high efficiency at output
units can be achieved. In addition, in wireless devices with communication media, the
most potent transmitters are not used; hence, the strength of identification is much lesser.
However, in the proposed method using deep learning, better optimization characteristics
are achieved. Therefore, the identification process is made much simpler. The robustness
characteristics of the proposed and existing method are provided in Figure 8.
Figure 8. Robustness characteristics.
From Figure 8, it can be observed that the strength of the proposed wearable devices
for identifying a particular infection is much higher compared to the existing method. To
verify the robustness characteristics, more iterations are considered from 10 to 100, where,
during each variation, the strength of identification changes between lower and upper
values. However, during these changes, the proposed method maintains tolerable limits,
as the threshold capacity of the device is predefined in the wearable device. However, in
the existing method [3], even if the low values are changed, tolerable limits are not main-
tained in the system. As a result of low tolerable conditions, the robustness of the existing
method remained at 5%, while the proposed method provides 1% robustness to the max-
imum extent. These comparison values are plotted concerning the simulation time stand-
ard in MATLAB, but the proposed wearable devices provide much low robustness in real-
time, as the loss factor is minimized.
5. Conclusions
Figure 8. Robustness characteristics.
5. Conclusions
Wearable Internet of Things (IoT) devices not directly inserted into any human body
system are used to examine the most crucial need in the current generation of networks
used for identifying various infections. However, because a communication unit is present,
all central servers in the system can be directly contacted by wearable devices, allowing
all specialists to provide medical information remotely. The design of wearable devices
must be distinctive, to identify every infection at early onset, where possible. For this
reason, mathematical design parameters are introduced in the proposed method to design a
wearable device capable of identifying parametric values. Additionally, with the designed
formulations, it is crucial to incorporate an optimization algorithm that offers successful
solutions under low-risk circumstances. As a result, a deep learning procedure is included
in our model. Thanks to the integrations above, all offloading requirements must be met to
convert standard devices to smart devices. In addition, a cloud-based data-collection unit
is present, where IoT device outputs are sent for decision-making. The proposed method’s
results are validated using five experimental scenarios, including distance measurements,
data waiting times, sensitivity levels, quality assessments, and loss assessments. The
proposed method offers the best results, by about 68 percentage points more than the
existing state-of-the-art method, according to the parametric output values for all the
test case verifications of designed wearable devices. By making the materials used lighter,
wearable device models with advanced designs can be possible in the future, and automated
optimization processes can be used to improve security and privacy in alternate ways.
Author Contributions:
Data curation: O.M.M. and H.M. (Hariprasath Manoharan); original draft
writing: H.M. (Hariprasath Manoharan) and S.S.; supervision: H.M. (Hana Mujlid) and S.S.; project
administration: S.S.; conceptualization: H.M. (Hariprasath Manoharan); methodology: H.M. (Hana
Mujlid) and S.S.; validation: O.M.M. and H.M. (Hariprasath Manoharan); visualization: O.M.M. and
H.M. (Hariprasath Manoharan); resources: H.M. (Hana Mujlid) and S.S.; review and editing: G.S.
and M.A.K.; funding acquisition: G.S. and M.A.K. All authors have read and agreed to the published
version of the manuscript.
Funding: This research received no external funding.
Diagnostics 2022,12, 2750 17 of 17
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement:
The data presented in this study are available on request from the
corresponding author. The data are not publicly available due to the program code transfer path.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Lan, L.; You, L.; Zhang, Z.; Fan, Z.; Zhao, W.; Zeng, N.; Chen, Y.; Zhou, X. Generative Adversarial Networks and Its Applications
in Biomedical Informatics. Front. Public Health 2020,8, 164. [CrossRef] [PubMed]
2.
Asthana, S.; Megahed, A.; Strong, R. A Recommendation System for Proactive Health Monitoring Using IoT and Wearable
Technologies. In Proceedings of the 2017 IEEE International Conference on AI & Mobile Services (AIMS), Honolulu, HI, USA,
25–30 June 2017; pp. 14–21.
3.
Iqbal, S.M.A.; Mahgoub, I.; Du, E.; Leavitt, M.A.; Asghar, W. Advances in healthcare wearable devices. NPJ Flex. Electron.
2021
,
5, 9. [CrossRef]
4.
Mohammadzadeh, N.; Gholamzadeh, M.; Saeedi, S.; Rezayi, S. The application of wearable smart sensors for monitoring the vital
signs of patients in epidemics: A systematic literature review. J. Ambient Intell. Humaniz. Comput. 2020. ahead of print.
5.
Manoharan, S.N.; Kumar, K.M.V.M.; Vadivelan, N. A Novel CNN-TLSTM Approach for Dengue Disease Identification and
Prevention using IoT-Fog Cloud Architecture. Neural Process. Lett. 2022. ahead of print. [CrossRef] [PubMed]
6.
Pal, D.; Triyason, T.; Varadarajan, V.; Zhang, X. Quality of experience evaluation of smart-wearables: A mathematical modelling
approach. In Proceedings of the 2019 IEEE 35th International Conference on Data Engineering Workshops (ICDEW), Macao,
China, 8–12 April 2019; pp. 74–80.
7.
Ramasamy, L.K.; Khan, F.; Shah, M.; Prasad, B.V.V.S.; Iwendi, C.; Biamba, C. Secure Smart Wearable Computing through Artificial
Intelligence-Enabled Internet of Things and Cyber-Physical Systems for Health Monitoring. Sensors 2022,22, 1076. [CrossRef]
8.
Ghazal, T.M.; Hasan, M.K.; Abdullah, S.N.H.; Abubakkar, K.A.; Afifi, M.A.M. IoMT-enabled fusion-based model to predict
posture for smart healthcare systems. Comput. Mater. Contin. 2022,71, 2579–2597. [CrossRef]
9. Ogura, Y.; Parsons, W.H.; Kamat, S.S. Cravatt, HHS Public Access. Physiol. Behav. 2017,176, 139–148.
10. Kshirsagar, P.R.; Manoharan, H.; Shitharth, S.; Alshareef, A.M.; Albishry, N.; Balachandran, P.K. Deep Learning Approaches for
Prognosis of Automated Skin Disease. Life 2022,12, 426. [CrossRef]
11.
Ashfaq, Z.; Rafay, A.; Mumtaz, R.; Hassan Zaidi, S.M.; Saleem, H.; Raza Zaidi, S.A.; Mumtaz, S.; Haque, A. A review of enabling
technologies for Internet of Medical Things (IoMT) Ecosystem. Ain Shams Eng. J. 2022,13, 101660. [CrossRef]
12.
Salehi, W.; Gupta, G.; Bhatia, S.; Koundal, D.; Mashat, A.; Belay, A. IoT-Based Wearable Devices for Patients Suffering from
Alzheimer Disease. Contrast Media Mol. Imaging 2022,2022, 3224939. [CrossRef]
13.
Ko, J.; Choi, Y.J.; Paul, R. Computation offloading technique for energy efficiency of smart devices. J. Cloud Comput.
2021
,
10, 44. [CrossRef]
14.
Mukhopadhyay, S.C.; Suryadevara, N.K.; Nag, A. Wearable Sensors for Healthcare: Fabrication to Application. Sensors
2022
,22,
5137. [CrossRef] [PubMed]
15.
Teekaraman, Y.; Manoharan, H.; Kirpichnikova, I. Abridged design with demand conventions for health care applications.
Comput. Electr. Eng. 2022,102, 108203. [CrossRef]
16.
Khajeh-Khalili, F.; Khosravi, Y. A novel wearable wideband antenna for application in wireless medical communication systems
with jeans substrate. J. Text. Inst. 2021,112, 1266–1272. [CrossRef]
17.
Cheng, Y.; Wang, K.; Xu, H.; Li, T.; Jin, Q.; Cui, D. Recent developments in sensors for wearable device applications. Anal. Bioanal.
Chem. 2021,413, 6037–6057. [CrossRef] [PubMed]
18.
Radoglou-Grammatikis, P.; Sarigiannidis, P.; Efstathopoulos, G.; Lagkas, T.; Fragulis, G.; Sarigiannidis, A. A Self-Learning
Approach for Detecting Intrusions in Healthcare Systems. In Proceedings of the ICC 2021—IEEE International Conference on
Communications, Montreal, QC, Canada, 14–23 June 2021.
19.
Nahavandi, D.; Alizadehsani, R.; Khosravi, A.; Acharya, U.R. Application of artificial intelligence in wearable devices: Opportu-
nities and challenges. Comput. Methods Programs Biomed. 2022,213, 106541. [CrossRef] [PubMed]
20.
Manoharan, H.; Selvarajan, S.; Yafoz, A.; Alterazi, H.A.; Chen, C. Deep Conviction Systems for Biomedical Applications Using
Intuiting Procedures with Cross Point Approach. Front. Public Health 2022,10, 1–14. [CrossRef] [PubMed]
21.
MacIa-Fernandez, G.; Garcia-Teodoro, P.; Diaz-Verdejo, J. Fraud in roaming scenarios: An overview. IEEE Wirel. Commun.
2009
,
16, 88–94. [CrossRef]
22. Ashok, R.L.; Agrawal, D.P. Next-Generation Wearable Networks. Computer 2003,36, 31–39. [CrossRef]
23.
Medeiros, A.; Leme, L.; Srivastava, G. An Introduction to Wearable Sensor Technology. In Intelligent Internet of Things for Healthcare
and Industry; Springer: Cham, Switzerland, 2022; pp. 189–198.
24.
Wang, W.; Chen, Q.; Yin, Z.; Srivastava, G.; Gadekallu, T.R.; Alsolami, F.; Su, C. Blockchain and PUF-based lightweight
authentication protocol for wireless medical sensor networks. IEEE Internet Things J. 2021,9, 8883–8891. [CrossRef]
