Hindawi
Journal of Healthcare Engineering
Volume 2022, Article ID 7713939, 16 pages
https://doi.org/10.1155/2022/7713939
Research Article
IoT-Enabled Framework for Early Detection and Prediction of
COVID-19 Suspects by Leveraging Machine Learning in Cloud
Mahmood Hussain Mir ,1 Sanjay Jamwal ,1 Abolfazl Mehbodniya ,2 Tanya Garg ,3
Ummer Iqbal ,4 and Issah Abubakari Samori 5
1Department of Computer Sciences, Baba Ghulam Shah Badshah University, Rajouri, Jammu and Kashmir 185234, India
2Department of Electronics and Communication Engineering, Kuwait College of Science and Technology (KCST), Doha Area,
7th Ring Road, Kuwait
3.apar Institute of Engineering and Technology, Patiala, Punjab, India
4National Institute of Technology Srinagar, Srinagar, J&K, India
5School of Engineering Sciences, University of Ghana, Accra, Ghana
Correspondence should be addressed to Mahmood Hussain Mir; mahmoodhussain@bgsbu.ac.in and Issah Abubakari Samori;
iasamori@st.ug.edu.gh
Received 13 January 2022; Revised 12 February 2022; Accepted 14 March 2022; Published 12 April 2022
Academic Editor: Suneet Kumar Gupta
Copyright © 2022MahmoodHussainMir et al.+is is an open access article distributed under the Creative CommonsAttribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
COVID-19 is the repugnant but the most searched word since its outbreak in November 2019 across the globe. +e world has to
battle with it until an effective solution is developed. Due to the advancement in mobile and sensor technology, it is possible to
come up with Internet of things-based healthcare systems. +ese novel healthcare systems can be proactive and preventive rather
than traditional reactive healthcare systems. +is article proposes a real-time IoT-enabled framework for the detection and
prediction of COVID-19 suspects in early stages, by collecting symptomatic data and analyzing the nature of the virus in a better
manner. +e framework computes the presence of COVID-19 virus by mining the health parameters collected in real time from
sensors and other IoTdevices. +e framework is comprised of four main components: user system or data collection center, data
analytic center, diagnostic system, and cloud system. To point out and detect the COVID-19 suspected in real time, this work
proposes the five machine learning techniques, namely support vector machine (SVM), decision tree, naı̈ve Bayes, logistic
regression, and neural network. In our proposed framework, the real and primary dataset collected from SKIMS, Srinagar, is used
to validate our work. +e experiment on the primary dataset was conducted using different machine learning techniques on
selected symptoms. +e efficiency of algorithms is calculated by computing the results of performance metrics such as accuracy,
precision, recall, F1 score, root-mean-square error, and area under the curve score. +e employed machine learning techniques
have shown the accuracy of above 95% on the primary symptomatic data. Based on the experiment conducted, the proposed
framework would be effective in the early identification and prediction of COVID-19 suspect realizing the nature of the disease in
better way.
1. Introduction threat across the globe, as of now it took away around 5.15
million lives. COVID-19 comes under the family of Coro-
COVID-19 is a repugnant word across the globe since its naviridae, which causes illness from common cold to more
breakout from Wuhan City of China in November 2019. severe diseases. In 2012, Saudi Arabia was epicenter for
COVID-19, the name given by the World Health Organi- MERS-CoV with 35% fatality rate [2]. In 2003, Southern
zation (WHO), initially erupted as an epidemic, but later China reported the SARS-CoV, which is also from a family
turned into a deadly pandemic [1]. In November 2021, the of the same virus. Later, both SARS-CoV and MERS-CoV
figures of COVID-19-confirmed cases exceeded 257.46 spread across the globe [3]. COVID-19 from its very in-
million with 3.7% mortality rate. COVID-19 spread the ception in November 2019 changed its physical and chemical
2 Journal of Healthcare Engineering
properties. +e novel strains of COVID-19 are more vul- and information syncing, and the IoTplays a notable role in
nerable and transferable with high risk of infection [4]. both of these areas [17, 18]. +e future predictions can be
WHO proclaimed the new COVID-19 pandemic on March made using AI and ML in IoT-based systems for predicting
11, 2020. To stop the spread of COVID-19, most of the the upcoming infection of coronavirus [19]. +e IoT can be
countries across the globe have shut down all the traffic used as a data source, and ML is used for data analytics to
including air, railways, and markets. Many countries have better further analyze the COVID-19 [20] to get better in-
also imposed restrictions or locked down the cities. sights. With the help of IoT, a centralized information
+e virus has wreaked havoc on the whole food chain, system can be created where all activities are stored elec-
revealing its fragility. Due to frontier closures, business tronically and can be accessed anywhere and anytime [21]. A
circumscriptions, and incarceration measures, the general vast number of people die because of lack or incorrect and
small-scale businesses, street vendors, vegetable growers, inappropriate knowledge about their health. +e use of IoT
and daily wagers were unable to access their local selling technology can quickly notify individuals’ health parameters
places, including obtaining inputs and selling their goods, through the deployed or wearable sensors [22]. +e IoT
disrupting national and global food supply networks, and technology can watch and capture the routine activities of an
restricting access to nutritious, safe, and diverse meals [5]. individual and can generate the necessary alerts if there is
From research perspective, COVID-19 is the most searched any critical health issue [23].
term on Internet in 2020. A lot of research related to
COVID-19 is currently going on throughout the globe [6].
Medical professionals are trying to come up with an antidote 1.1. Motivation and Contribution. COVID-19 has taken
that can prevent corona infection. From the perspective of millions of lives since its outbreak started inWuhan, a city in
Internet of things, a vast research is being conducted on the China, from the month of November 2019. A lot of research
impact of IoT technology to tackle the COVID-19 epidemic is going on across the globe to combat this pandemic, but the
[7]. Computer scientists on the other hand are trying to strategies and procedures for analyzing and predicting the
develop models that can detect and prevent the infection. virus are still in its infancy. As the pandemic spread around
+e traditional healthcare system is not sufficient enough to the globe, healthcare systems collapsed due to the un-
handle the current global prevalent situation. Presently, the availability of smart diagnostic systems. Due to the fast
only way to avoid COVID-19 infection is to follow the SOPs transmission of COVID-19 from person to person, an IoT-
and get vaccinated with immunity boosters. +e advance- based system will help in predicting the onset of infection in
ment and increase in mobile technology such as sensors, a real time, thus in turn help in the prevention of this deadly
smart devices, and other wearables mingled in the healthcare disease. Healthcare system across the globe is poor due to the
system greatly impact our daily lives [8]. Nowadays, IoT is lack of integration of technology. +e IoT can help the
mingled in every field, with the ability to communicate from healthcare system to automate many sectors to eliminate the
anywhere, anytime [9] round the clock. New and advanced errors made by humans. On the other hand, machine
powerful devices for monitoring individuals’ health came learning can be used for analysis purposes to get better
due to IoT [10]. IoT is the integration of physical devices insights and understand the nature of disease. +e inte-
with communication technologies capable of connecting gration of both the technologies such as IoT, ML, and AI can
through the Internet. +e real-time health parameters are revolutionize the modern healthcare system. By incorpo-
taken from deployed sensors to provide the current status of rating machine learning in the domain of health care, most
patients [11]. In the current era, mobile phones have inbuilt of the things can be achieved such as maintaining accurate
onboard sensors that can capture the real-time parameters of data, personalized healthcare facilities, and predictive ana-
patients. +e various security mechanisms are employed in lytics. IoTcan mainly be used as for sensing the environment
sending and receiving data from these smart applications and actuating accordingly, but machine learning is for high-
[12]. Smartphones can be used as input devices such as end analytics. +e proposed framework based upon the
sensing, storing, and computing the results [13]. By the use machine learning and IoT will act in a proactive and pre-
of technology, it is possible to detect the COVID-19 suspects ventive manner rather than in reactive manner as used in
in early stages to eliminate the spread of infection. Tracking traditional approaches of prevention.
and quarantining of COVID-19-positive and COVID-19- +is article proposes a layered architecture and early
suspected cases can be tracked and monitored with the help detection and monitoring system of COVID-19 suspects.
of onboard mobile phone sensors [14] and by the wireless Using IoT devices, a real-time symptomatic data are col-
sensor technology [15] (WSN). Integration of IoTwith other lected to identify COVID-19 potential cases. By deploying
potential technologies such as machine learning (ML) and IoT-based sensors, there are mainly three potential advan-
artificial intelligence (AI) can revolutionize the healthcare tages: firstly, continuous monitoring anytime and anywhere.
sector in near future [16]. As a result, in the face of the Second, frequent symptomatic parameters are collected on
pandemic, artificial intelligence (AI) and machine learning regular basis. +irdly, the regular symptomatic data are
(ML) created new potential options for successful therapy. collected in a particular time frame. To detect the COVID-19
AI and machine learning can be used in the discovery of new suspect at an early phase, a set of parameters (symptoms) are
drugs, the development of accurate diagnostic processes, and required for effective results, which is impossible in a single
the prediction of disease vulnerabilities. +ese potential visit to the clinic. To overcome the cons of the traditional
areas are strongly reliant on real-time patient monitoring healthcare system, a novel COVID-19 detection and
Journal of Healthcare Engineering 3
prediction and monitoring system in early stage are pro- low root-mean-square error [27]. In this article, the authors
posed. +e proposed framework contributes by 1. early have made the step-by-step review of the artificial intelli-
detection of COVID-19 suspects, 2. analyzing the symp- gence in healthcare domain.+e AI is comprised of machine
tomatic collected data using machine learning techniques, 3. learning and deep learning for prearranged datasets, whereas
disease diagnosis (whether COVID-19-positive or COVID- text mining and natural language processing are for unar-
19-negative), and 4. maintaining health record of patients ranged datasets. +e authors have highlighted the challenges
for future use. +e main aim of the proposed system is to and research opportunities by integrating AI in healthcare
eliminate the expansion of the coronavirus infection and sector. +e authors have discussed in a long the technologies
detect the COVID-19 in the early phase, and the disease can that can combat the pandemic [28].+e authors in this paper
be better understood from the collected data by further have developed machine learning-based efficient automatic
analysis. disease model based on android application. +e model has
Lastly, our proposed framework has been tested through been tested on three different diseases such as COVID-19,
a novel dataset collected from SKIMS, Srinagar. Distinct ML diabetes, and cardiovascular disease. +e authors have used
algorithms have been employed on the dataset to validate the logistic regression algorithm for prediction and have
system. Our system has achieved above 95% accuracy. +e comparative analysis. Industry 4.0 has revolutionized the
proposed system is cross-validated using various perfor- world with the advancements in ICTs in easing human lives
mance matrices such as accuracy, precision, recall, F1 score, [29]. Internet of things (IoT) is one of the main components
root-mean-square error, and area under the curve. of Industry 4.0 that has changed the way of thinking. IoT is
+e proposed framework comprised of four main internetwork of physical objects embedded with sensors,
components: 1. user system: in which sensors are collecting communication technologies, processing abilities, and other
real-time symptomatic data, 2. data analytic center: various technologies [30]. COVID-19 is influenza-type disease,
machine learning algorithms are employed to collect data, 3. which causes the infection in respiratory system with
diagnostic system: healthcare (physicians) experts check the symptoms such as fever, cough, runny nose, and breath-
calculated parameters, and 4. cloud system. +e aim of the lessness. It spread faster from person to person by coming in
framework is to eliminate the death rate by early detection contact, so predicting the spread of infection is challenging.
and eliminate the spread of coronavirus infection.+e paper +e authors proposed a model to diagnose the COVID-19
uses five (5) machine learning algorithms namely SVM, infection. +ree types of techniques have been tested on the
decision tree, näıve Bayes, logistic regression, and neural Kaggle dataset such as linear regression, multilayer per-
network. +e paper tests the proposed framework by ceptron, and vector autoregression. Reference [31] has made
experimenting the ML techniques on novel dataset. +e the systematic review of healthcare technologies such as IoT,
experimental results have shown that these algorithms have big data, and cloud computing with respect to Industry 4.0.
achieved above 95% accuracy. A lot of literature studies have been surveyed to discuss the
+e rest of the paper is organized as follows: Section 2 main technologies and application of IoT in healthcare.
elaborates the detailed relevant literature. Section 3 gives the Riazul Islam et al. presents a comprehensive survey of IoT
detailed insights of the proposed system. Section 4 discusses in health care. +e authors have reviewed a state-of-art
the experimental setup. Section 5 provides the detection and literature about technologies, architectures, and applications
prediction model for potential COVID-19-suspected cases of Internet of things in healthcare. Security models have also
by employing machine learning techniques. Section 6 is the been discussed and presented as a security model for IoT
results and discussion of the proposed work. Lastly, Section 7 healthcare [33]. +e authors have discussed the possibilities
concludes the work. of integration of artificial intelligence with the wireless
technologies to combat in pandemic situations. In this study,
2. Related Work the authors have proposed an ensemble machine learning
model, i.e., random forest algorithm to predict the severity of
AI and machine learning have opened doors for large array COVID-19 patients taking under several parameters. +e
of applications in the medical industry, including statistical proposed model has performed well in almost all perfor-
data prediction and classification [24]. BlueDot Toronto, for mance measures such as accuracy, F1 score, precision, and
instance, established the first risk-based technique for rec- recall. +e proposed algorithm is compared with other al-
ognizing the SARS-CoV-2 epidemic, which was developed gorithms such as SVM, decision tree, logistic regression, and
by IoT 2020 by infectious disease professionals to investigate naı̈ve Bayes. +e proposed algorithm surpasses all the al-
new solutions for mitigating the initial SARS pandemic. gorithms in terms of performance measures. +e proposed
BlueDot’s previous SARS research was utilized to include algorithm achieved an accuracy of 94%, F1 score of 0.86,
advanced technologies in this impressive demonstration of precision of 1.0, and recall of 0.75. Reference [34] has
AI and ML in forecasting illness outbreaks [25]. In [26], the proposed a cloud-IoT-based framework for student health
authors have developed machine learning-based framework monitoring. +e proposed framework predicts the level of
for diabetes prediction and named it as intelligent diabetes disease by measuring temporal measurements by collecting
mellitus prediction framework (IDMPF). +e authors pro- data from medical IoT devices. +e authors of their study
posed three machine learning techniques to predict diabetes have used a dataset of 182 students to test the proposed
that are as support vector machine, random forest, and framework. Various machine learning algorithms have been
decision tree. +ey have achieved an accuracy of 83% with applied and validated using k-cross-validationmethods [35].
4 Journal of Healthcare Engineering
A lot of literature studies have been reviewed, and the image processing, data analytics, text mining, and natural
potential application of IoT has been discussed. +e article language processing are some areas that are discussed in this
came under COVID-19 solutions with current applications article. A detailed overview of open COVID-19 datasets is
of IoT such as smart transportation, ambient living, and publicly accessible for research purposes. +e authors have
smart city [36]. +e authors have remote asthma patient also discussed the future directions of potential areas of AI
monitoring system based on IoT technology. +e moni- that can fight against COVID-19. Siriwardhana et al. [44]
toring system is comprised of sensors, android application, present the power of 5G and IoT to combat COVID-19. +e
and website. +e sensors are collecting vital parameters such authors have discussed several use cases of these technol-
as blood pressure and glucose level, and the model was tested ogies that can provide innovative solutions such as contact
on some patients [37]. Internet of things is a disruptive tracing, telehealth, and education [45]. +e present situation
technology that can renovate the healthcare system. +e has opened the doors for creating new avenues in our daily
authors have made good efforts on how IoT can be lives. +e authors have a lot of literature studies about the
implemented to tackle COVID-19. +ey have given a brief COVID-19 solutions and have identified seven potential
insight of various IoT technologies that can be used during applications useful during pandemic [46]. +e authors have
the COVID-19 pandemic [38]. +e vaccine is developed by reviewed the literature on machine learning techniques and
different companies such as BioNTech, Pfizer, and Moderna IoT in combating COVID-19 pandemic. +e medical
in India. +e vaccines have different effects on the people methods are time-consuming and costly such as RT-PCR
based on demographic factors. +e researchers in this study and CT (chest) and are putting burden to technologists and
have analyzed the data collected from vaccine companies to radiologists. AI is a potential technology that can eliminate
predict the viable persons based on some variables. +e the cost and time to combat the COVID-19 pandemic. +e
variables are age, gender, and others such as state of living. authors have also discussed the challenges of IoT and ML in
Based on these parameters, the researchers are predicting the fighting the COVID-19 pandemic [47]. COVID-19 has af-
best manufacturer for that person. +e researchers have fected almost each and every field. In this article, the authors
employed different machine learning algorithms such as have discussed the literature on IoT and ML to prevent and
logistic regression, decision tree, random forest, and Ada- diagnose the COVID-19 pandemic. +e authors have
Boost. +e performance measures of these algorithms are explained the various machine learning techniques for
contrasted in terms of accuracy.+e AdaBoost has surpassed classification and clustering for COVID-19. Reference [48]
all and achieved 98.1% accuracy, random forest has 97.8% of has highlighted that as a consequence of the COVID-19
accuracy, and decision tree and logistic regression are at the problem, several enterprises have closed, and many
same place with 97.3% of accuracy [39]. IoT can be used to manufacturing and small merchants will go out of business.
eliminate spread of COVID-19. +is technology helps in +ey must deal with a myriad of difficulties, such as cost
providing more user satisfaction by properly monitoring containment and worker sanitation. Several strategies for
COVID-19 patients. +e authors have explored twelve coping with the pandemic crisis have been presented, with
potential areas of IoT to combat COVID-19. IoT is helpful in IR 4.0 playing an important role. Reference [49] has pro-
identifying the symptoms of COVID-19 suspects to provide posed a hybrid model to predict the mortality rate on the
better treatments [40]. A cloud-IoT-based platform for India in future. +ey have used statistical neural network
disease diagnosis has been presented by the authors. +e (SNN) and nonlinear autoregressive neural network (NAR-
proposed paradigm forecasts the severity of a potential NN)-based models to improve the prediction accuracy. +e
disease. +e suggested framework has been tested using the results are compared with SNN-based models such as
UCI dataset. To estimate the severity of disease, various probabilistic neural network (PNN), radial basis function
machine learning classification techniques were employed to neural network (RBFNN), and generalized regression neural
the obtained data.+e accuracy, sensitivity, specificity, and F network (GRNN). +e performance of the models is mea-
measure were used to calculate the findings [41]. According sured using root-mean-square error (RMSE) and R (cor-
to the report, the employment of robotics, IoT, and other relation coefficient). +e hybrid model of PNN and RBFNN
related innovations has expanded rapidly as a result of the performed better than all [50]. +e authors have suggested
rise of Industry 4.0. +e Internet of things (IoT) is a strong the IoT-based identification and control system in real time.
solution for a wide range of real-time issues, thanks to the +e system identifies the potential cases in early stages and
sensors that make it possible. IoTacts as a crucial enabler for tracks their clinical measures. +e proposed framework has
Industry 4.0 through device connectivity, enabling better five main components: data collection, quarantine center,
management, customized service, and efficient operation processing unit, cloud computing, and visualization of data
[42]. +e authors have developed a cloud-based disease to healthcare professionals. +e authors have employed
forecast and diagnostic system using various algorithms.+e various machine learning techniques to detect COVID-19
input is collected from IoT wearable devices and then suspects [51]. IoT is a vital technology that has the potential
transferred these signals to a server using Internet. +e to combat during pandemic such as COVID-19.+e authors
authors first create the feature set from collected data using in this paper have proposed a four-layer model to predict
the proposed hybrid decision-making approach.+e authors potential cases of COVID-19. +e model has four compo-
have also proposed IoT-based framework with flow of in- nents: data acquisition, data aggregation, machine intelli-
structions in their research paper. Reference [43] discussed a gence, and services. +e model is validated using voice data
lot of AI techniques used to tackle COVID-19. Medical [52]. +e authors have surveyed a lot of literature studies of
Journal of Healthcare Engineering 5
IoT technologies used in tracing, tracking, and spread of
COVID-19. +e authors have highlighted the architectures CLOUD LAYER
and also future directions of IoT implementations [10]. +e
authors in this article have highlighted the applications of
IoT that can be used in combating COVID-19. +e authors ANALYSIS LAYER
have proposed a real-time identification and monitoring
system for COVID-19. +e model is divided into four
components based on cloud technology: the collection of SENSING LAYER
symptomatic data, health center, data warehouse, and health
professionals. +e authors have tested the framework using Figure 1: Proposed 3-layer architecture.
machine learning models, and random forest has shown the
best results. sensor, and inertia sensor. +e other relevant parameters are
collected from user through applications such as travel
3. Proposed IoT Framework history through smartphones history. +ese sensors are
connected with IoTgateway to communicate the sensed data
3.1. Proposed Architecture. +is section discusses the IoT- through Internet. Sensors are battery-powered so they are
cloud architecture of the proposed system, diagrammatically not directly communicating to Internet.+e communication
presented in Figure 1: proposed 3-layer architecture. +e technology used by sensors to communicate with gateway is
proposed layered architecture is based on standard IoT ar- low-powered technology such as BLE, infrared, and Wi-Fi.
chitecture, and it has three layers: sensing layer, analysis layer, +e gateway uses Wi-Fi, mobile networks, 3G, 4G, 5G, etc.,
and cloud layer. +e sensing layer or perception layer is ac- to communicate with the cloud system.
countable for the collection of symptoms from the suspected Data Analytic Center: this component is responsible for
persons through various deployed sensors, wearables, and IoT data analysis and hosting of machine learning algorithms.
devices. +ere are various types of electronic digital sensors On the basis of collected symptoms accessed from personal
such as temperature sensor, audio-based sensors,motion-based health records of cloud system, prediction is made whether a
sensors, heart rate sensor, O2 sensor, and other biosensors such person is COVID-suspected or not. +e results are then
as ECG and EEG. Other information such as travel history and generated and updated in cloud accordingly. As it is con-
other parameters are collected with the help of applications. tinuously updating the personal health records, the machine
+e sensing layer sends this collected information to the layer learning models are updating also with the help of new
above it called as the analysis layer. +e analysis layer is re- analysis made by data analytic module.
sponsible for doing analysis of data received from the sensing Medical Laboratory and Diagnostic System: this module
layer. Numerous machine learning models are deployed in this is comprised of health physicians and medical laboratories.
layer for getting better insights from data. +e prediction of +e suspected first are sent for laboratory test (RT-PCR/
suspected cases is made based on symptoms of a person of RAT), and if they are found positive, they are checked by
whether a suspected is COVID-19-positive or not. +e re- medical physicians for health checkup. +e clinical inves-
sultant data are then sent to the cloud layer for other services. tigations are made based on patient’s symptoms received
+e third layer of the architecture is the cloud layer, which is from cloud system. +is proposed model can predict and
responsible for storing the data. Healthcare professionals can eliminate the further spread of COVID-19-suspected cases.
then use the stored data for further analysis. +e data are used Cloud System: cloud computing is buzz term for last two
to update machine learning models for deriving more accurate decades, in which everything is in logical way in a centralized
results. system known as cloud. On-demand services are provided
such as storage, databases, and computing resources in a
3.2. Proposed Framework. +is section discusses the pro- cloud computing environment. In our case, all types of
posed IoT-based framework to identify and predict COVID- services such as storage and computing resources are taken
19 suspects in early stages. +is framework is also used to from the cloud environment. +e data sensed by the sensing
eliminate the further spread of infection and get better layer are communicated via communication networks to
insights of the disease for future perspective. Figure 2: a cloud for storage purposes, updating personal health rec-
conceptual framework for early detection and prediction of ords, and communicating with other components.
COVID-19 suspect, shows the proposed model of the sys-
tem. +e framework has mainly consisted of three main
modules with respect to the proposed three-layer archi- 3.3. Flowchart of Proposed Framework. +e flow of frame-
tecture: user system, data analysis system, and cloud system. work is described in Figure 3: data flow of proposed
User System: the main objective of this module is to framework, and the steps are described as follows:
sense real-time data with the help of sensors and wearables. (1) +e system collects data from sensors and wearables
+e collected symptom data are fever, cough, fatigue, rhi- deployed through body area network (BAN). +e
nitis, breathlessness, myalgia, oxygen saturation, travel symptoms such as cough, rhinitis, sore throat,
history, blood pressure, etc. +ere are several sensors such as breathlessness, O2 saturation, blood pressure, and
temperature sensor, O2 sensor, motion sensor, proximity other related information through smartphone are
6 Journal of Healthcare Engineering
Users CLOUDSYSTEM
INTERNET
Doctors
Users
Diagnosis System
Data Analytics Center
Figure 2: A conceptual framework for early detection and prediction of COVID-19 suspect.
2. (i) Analysing and
Monitoring of Data
collected from Sensors
and other Devices 3. (i) Sharing Data withHealthcare Centers and
1. Sensing of Data from differnet Healthcare
sources such as Sensors, Data Analysis Professionals
Wearables, Smartphone etc. such
as Fever, Shortness of Breath, 3. (ii) Sending Patient
Cough, Fatigue, Travel History Data for Lab Test
etc. Uploading Suspected RTPCR/RAT
Disease Diagonosis
2. (ii) Running Machine
Learning Algorithms such
as SVM, Naive Bayes,
Decision Tree etc. 3. (iii) If Positive
Symptom Data Sending the Patient2. (iii) Identifying and for Consultation
Predicting of COVID-19
suspected cases
Figure 3: Data flow of the proposed framework.
collected in real time.+e collected data are then sent disease. +e COVID-19-suspected cases are pre-
for analysis. dicted and identified using machine learning models.
(2) +e uploaded data from step 1 is then analyzed for (3) If a person is COVID-19-suspected, they will be sent
possible COVID-19 infection. +e machine learning for clinical laboratory test (RAT/RT-PCR) for in-
models are then applied to the collected data and vestigation. If suspected is COVID-19-positive, they
obtained the results. +e machine learning models will be sent to medical physician for checkup. +e
are continuously updating with the real-time data to confirmed positives can then be secluded, and all
derive more accurate results. Further, the results are other previous contacts will also be isolated to
seen by medical physicians to better understand the eliminate further spread of infection.
USER SYSTEM
SENSORS
Sensing Layer
IoT Gateway
Journal of Healthcare Engineering 7
4. Experimental Setup 4.2. Preprocessing, Feature Selection, and Normalization.
+e collected data from the SKIMS Institute are pre-
4.1. Data Collection. COVID-19 was declared a pandemic processed as follows: in the first phase, the more relevant
on March 11, 2020, by the World Health Organization. +e attributes or features have been selected. +e common
disease is new in nature and RNA-based and continuously features such as fever, cough, rhinitis, sore throat, and fa-
changes its properties. Due to these unpredictable proper- tigue have been selected to form a dataset. +e other less
ties, it is hard to derive any concrete solution. +e re- potential features such as hemoglobin, blood group,
searchers and hospitals give open access to data regarding comorbidities, anosmia, pulse, and BP have been discarded.
the confirmed cases.+e unpredictable and unknown nature Some of the attributes were merged such as loss of appetite
of the disease made it tough to develop any remedy or with anorexia, because of synonymity of words. After dis-
medicine. Researchers and academicians are trying to de- carding and merging process, less than 25 features were
velop a vaccine and a solution that can combat COVID-19. selected.+e second phase is preprocessing of data, in which
+e World Health Organization (WHO) and medical or- each column is checked for value. +ere are some missing
ganizations made it possible for everyone to contribute to or values for many of the cases written in the database. To
provide a solution to the COVID-19 pandemic. Researchers overcome that, some of the columns and rows were elim-
from different domains are trying their best to efficiently inated. Like values of BP, pulse was missing in most of the
solve the pandemic. Since the academic fraternity has no cases so these columns were deleted. Likewise, there were
prior experience of a pandemic such as COVID-19, none of some missing values in many rows; many rows were deleted
the solutions is a holistic working solution. As this has to overcome that. Lastly, our dataset was reduced to 6015
become an open challenge, the ongoing research is available rows and 21 columns as described in Table 2: selected
on different websites such as Google Cloud, NIH, COVID- symptoms of patients.
19 Data Repository, and other international and national Normalization is another important step to follow after
institutes. +e available public datasets are simple metadata finalizing the attributes of a dataset. Most of the attributes
or confirmed cases of different countries, by which a con- were categorical in nature such as travel history, residence,
crete solution cannot be drawn.+e available datasets do not cough, and sore throat. Some of the attributes were nu-
include all the information about patient’s symptoms be- merical such as fever, pulse, and oxygen saturation. So, to
cause of the novelty of the virus. +e available data are take the dataset into one form, the normalization is needed.
inadequate and insufficient for the use by machine learning In our case, most of the attributes have categorical value, so
algorithms. +is research aims to develop an IoT-cloud- other attributes are transformed into categorical value.
based system that can predict the COVID-19 suspects based Suppose if fever is above normal range, it is represented by 1,
on patient symptoms. +e actual dataset has been collected otherwise, 0. Similarly, all other attributes are converted to
from the Sher-I-Kashmir Institute of Medical Sciences categorical value to normalize the dataset. Our dataset is a
(SKIMS), Srinagar, Jammu and Kashmir, India, collabo- collection of rows and columns, in which each column
rating with the doctors. +e SKIMS is a renowned Medical represents a binary feature, either 1 or 0. +e value 1 of a
Institute of Jammu and Kashmir, India. During the pan- feature represents the presence of a symptom, and 0 feature
demic, they have received scores of COVID-19-positive represents the absence of that very symptom. Table 3: at-
patients for medical facilities. +e SKIMS Institute has made tributes of dataset, displays the attributes of dataset finalized
a separate temporary COVID-19 department. Before after the above steps and used during the work.
starting our work, a round table meeting was held with a
team of doctors to discuss the possible symptoms of
COVID-19 patients. +e symptoms of the COVID-19 pa- 4.3. Detection and Prediction of COVID-19 Potential Suspect.
tients were already published on various websites; in par- Machine learning (ML) is a type of artificial intelligence and
ticular, the set of primary symptoms given by WHO and subfield of computer science by which machines are learning
CDC on their websites are as follows: fever, cough, fatigue, without being explicitly programmed.ML is categorized into
runny nose, breathlessness, etc. +e dataset attributes three main categories: supervised learning, unsupervised
(symptoms) were finalized after consulting a group of senior learning, and reinforcement learning. In ML, a learning
doctors from the COVID-19 department of the institute. algorithm takes input from a set of variables known as a
Finally, the proforma of symptoms has been drafted to training set. +e training set of input values together with
collect those from COVID-19 OPD clinic and in-patients. target labels known as class labels is called supervised
+e list of attributes or symptoms is given in Table 1: col- learning. +e class labels are unknown in unsupervised
lected symptoms of patients. learning, and reinforcement learning means learning fol-
+ere are some other attributes such as travel history, lowing the action taken for a given situation. Since our
whether a patient is having any other diseases or not, such dataset is labelled, our focus will be on supervised learning.
as diabetes, kidney, and heart, blood group, hemoglobin, +e preprocessed dataset developed in the previous section
headache, anosmia, pulse, BP, respiratory rate, and is used to build a predictionmodel to identify the COVID-19
temperature. +e data of these attributes were either in- suspects. +e function of this model is to predict the possible
adequate or insufficient to take them as attribute.+us, the COVID-19 suspect by analyzing the symptoms of a person.
data preprocessing and feature selection must be Various ML algorithms have been employed on the dataset
performed. to classify them into either positive or negative. Depending
8 Journal of Healthcare Engineering
Table 1: Collected symptoms of patients.
Collected symptoms
MRD no. Date Age Sex Residence Contact history Fever Cough
Rhinitis Sore throat Shortness of breath Myalgia Fatigue Loss of appetite Loss of taste Vomiting
Nausea Diarrhea O2 saturation Hypertension Kidney disease Heart disease Liver disease Chest disease
Severity of illness Headache Body pain Anxiety Travel history Survival Test type Test result
Table 2: Selected symptoms of patients.
Selected symptoms
Age Sex Contact history Fever Cough Rhinitis Sore throat Shortness of breath
Myalgia Fatigue Loss of appetite Loss of taste Vomiting Nausea Diarrhea O2 saturation
Hypertension Other diseases Travel history Anxiety Chest pain Severity of illness Survival Test result
Table 3: Attributes of dataset.
Symptoms used for analysis
Travel history Contact history Fever Cough Rhinitis Sore throat Shortness of breath O2 saturation
Myalgia Fatigue Loss of appetite Loss of taste Vomiting Nausea Diarrhea Chest pain
Sex Other diseases Severity of illness Survival Test result
on the working, there are different categories of supervised probability of each class label. +en, this is used to
machine learning algorithms, such as regression-based: lo- assign the class label in the coming instance. MNB is
gistic regression, function-based: support vector machine, an extended version of NB that uses two or more NB
Bayes-based: naı̈ve Bayes, tree-based: decision tree, and variants. MNB uses the concept of term frequency to
meta-based: neural network. In this work, various machine compute maximum likelihood from the training data
learning techniques, such as SVM, decision tree, näıve Bayes, based on conditional probability.
logistic regression, and neural network, are used while (4) Logistic Regression: LG is a supervised machine
performing the task. learning technique borrowed from statistics. A
(1) Support Vector Machine: SVM is a supervised probabilistic model uses a logistic function to de-
machine learning classification technique. It takes termine the binary variable. Mathematically, a lo-
predefined set of input training examples with a gistic function is having dependent variable with two
given class label (i.e., positive (1) or negative (0)) as possible values, such as true or false in case of
input. SVM is a function-based learning algorithm COVID-19.
that divides the instances of each class with the (5) Neural Network: NN is also known as artificial NN
hyperplane. +e trained model is then used to (ANN) and is nature-inspired machine learning
predict the label for any new input. In our case, the technique. ANN is a meta-classifier-based ML
hyperplane is trained based on a patient’s symptoms technique that mimics how biological neurons are
with the given class label, either COVID-19-positive sending signals to one another. NN takes different
or COVID-19-negative. inputs i.e., neurons, and outputs one single output.
(2) Decision Tree: DT is a supervised machine learning NN is also known as multilayer perceptron because
technique. It takes a set of predefined training data many layers are in between, i.e., hidden layers.
with a given class label as input. DT is a tree-based
learning algorithm with three types of nodes: root
node, leaf nodes, and decision node. +e leaf node 5. Results and Discussion
exemplifies the class label, and the decision node Performance Evaluation: the performance evaluation of the
exemplifies the decision to make. DT normally fol- used machine learning algorithms is measured by six dif-
lows the disjunctive normal form (sum of product) ferent measures. +e six measures are accuracy, precision,
to form a tree. It uses many sub-algorithms and recall, F1 score, RMSE, and AUC score. +ese six measures
follows criteria such as information gain, entropy, were validated using confusion matrix and cross-validation
Gini index, and gain ratio, also known as vital methods.
function. Confusion Matrix: the visualization of performance of
(3) Multinomial Näıve Bayes: NB is a supervised ma- binary supervised machine learning algorithm is done by
chine learning technique based on the Bayes theo- creating a 2× 2 matrix. +e column represents the actual
rem, i.e., follows a probabilistic approach. For a given class, and the row represents the predicted or computed
set of training data with predefined labels, it com- class. +e matrix representation of 2× 2 confusion matrix is
putes model parameters by calculating the given in Table 4: confusion matrix.
Journal of Healthcare Engineering 9
Table 4: Confusion matrix. quadrant is an intersection of actual class and predicted
True positive (TP) False negative (FN) class. In our proposed system, the hold-out method is used,
False positive (FP) True negative (TN) in which the dataset is divided into training set and testing
set. +e dataset of 6015 rows is divided in the ratio of 70 : 30,
70% for training and 30% for testing, that is, 4210 rows for
True Positive: in this predictive model, the number of training and 1805 rows for testing. +e dataset is shuffled to
instances that were as positive is labelled as positive, and in eliminate the biases, so that the proposedmodel will perform
actual, they are positive. In a true positive result, the persons well in all situations. In our proposed system, the decision
that do have COVID-19 disease are predicted as positive. tree has shown best results in terms of false negatives, i.e.,
True Negative: in this, the model has classified the in- type II error. +e decision tree has ten false negatives from
stances as negative using predictive model, and in actual, the rest of the proposed machine learning techniques. +e
they are also negative. For example, in case of COVID-19, second place has naı̈ve Bayes algorithm with twelve false
the persons that do not have COVID are predicted by model negatives, and the third logistic regression, fourth neural
as negative. network, and last place have a support vector machine. From
False Positive: the model has classified some instances as this discussion, the proposed decision tree model has per-
positive using a predictive model, and in actual, they are formed well and it can still be enhanced with the data to
negative. In a false positive result, the persons that do not minimize the false negatives further.
have COVID-19 disease are predicted as positive. It is also Cross-Validation: it is a statistical technique used to
known as type I error. measure the performance of machine learning classification
False Negative: in this, the model has classified some techniques by splitting the training data into two sets. One
instances as negative using a predictive model, but in actual, set that is usually more than half is used for training, and the
they are positive. For example, in case of COVID-19, a rest of the data are for testing. +e seventy (70) percent is
person having COVID has shown not COVID by our model. used for training in our model, and thirty (30) percent is
It is known as type II error. used for testing. Each of the six performance measures
After applying machine learning techniques on the novel (accuracy, precision, recall, F1 score, root-mean-square
dataset, the resulted confusion matrices of applied machine error, and area under curve score) is calculated for all al-
learning algorithms are given in Figure 4: confusionmatrices gorithms and summarized in Table 6: summary of results of
of applied machine learning techniques (a, b, c, d, e). Di- different performance measures of applied machine learning
agonal elements represent good scores, and other (non- techniques.
diagonal) represent bad scores. +e results generated different performance measures
+e results generated in the confusionmatrices above are from the novel dataset, in which SVM has achieved the
summarized in Table 5: summary of results of confusion lowest of 97% and the rest of the algorithms have achieved
matrices of different applied algorithms. It is clearly visible 98% of accuracy. In terms of precision, the decision tree has
from the table that the experimentation has been performed achieved the highest of 99% and the rest of the algorithms
on the balanced data that remove the possibility of high bias have achieved 98%. +e decision tree has achieved 99%,
or variance. A simple look at the value of TP, TN, FP, and FN naı̈ve Bayes and neural network obtained 98%, and SVM and
tells the whole story about the classification results. In case of logistic regression have achieved 97% of recall.+e lowest F1
disease prediction, a classifier should have a smaller number score of 97% is achieved by SVM, 99% is achieved by the
of false negatives as cost is associated with the false negatives. decision tree, and the rest of the algorithms achieved 98%. In
Suppose in case of COVID-19 prediction, if the classifier has terms of AUC score, 97% is achieved by SVM and the rest
predicted any suspected falsely as COVID-19-negative, it have achieved 98%. +e RMSE should be low, DT and NN
will infect others. Otherwise, if the classifier has predicted have achieved 0.12, NB and LR have achieved 0.13, and SVM
any value as falsely positive, it will not infect others. On has 0.15. DT and NN have good value in terms of RMSE.
comparing the algorithms based upon the false negatives Our domain is health care, so the proposedmodel should
generated, it has been found that the decision tree performed have good score in all performance measures. +e proposed
better than the rest of the algorithms as fewer entries have model is to detect and predict COVID-19 suspect in early
been falsely predicted as negative. As in the above defini- stage to eliminate the spread and mortality rate of the in-
tions, there are two types of errors: type I error and type II fection. In this case, the recall of proposed technique should
error. Both the errors are not good for developed model, but be good so that the best can be achieved. COVID-19 is the
in case of disease the type II error is of main concern. repugnant term, so as positive fromNovember 2019. Corona
Suppose in case of COVID-19, if our model will drop a disease spreads from humans to humans by touching the
person in class of FN, it is type II error and it will infect the infectious person and by different ways. If the proposed
others. So, in case of disease the model should have low type model predicts a person falsely positive, it will not affect the
II error; otherwise, it will make huge cost to our proposed performance of model in our case. If a model detects a
model. person falsely negative, it will infect many, and it is not an
From Table 5, the values are clearly shown against each effective model. In technical terms, when a cost is associated
intersection point of the matrix presented in Figure 4. +e with false negative, recall is the best measure to check the
matrix is divided into four binary classifications; each model.
10 Journal of Healthcare Engineering
Confusion matrix of SVM Confusion matrix of DT
900
800 800
0 902 2 700 0 888 16 700
600 600
500 500
400 400
300 300
1 40 861 200 1 10 891 200
100 100
0 1 0 1
Predicted label Predicted label
(a) (b)
Confusion matrix of MNB Confusion matrix of LR
800 800
0 884 20 700 0 890 14 700
600 600
500 500
400 400
300 300
1 12 889 200 1 17 884 200
100 100
0 1 0 1
Predicted label Predicted label
(c) (d)
Confusion matrix of NN
800
0 894 10 700
600
500
400
300
1 19 882 200
100
0 1
Predicted label
(e)
Figure 4: Confusion matrices of applied machine learning techniques. (a) Support vector machine. (b) Decision tree. (c) Naı̈ve Bayes.
(d) Logistic regression. (e) Neural network.
Table 5: Summary of results of confusion matrices of different applied algorithms.
ML algorithm True positive (TP) True negative (TN) False positive (FP) False negative (FN) Support
Support vector machine 861 902 2 40 1805
Decision tree 891 888 16 10 1805
Naı̈ve Bayes 889 884 20 12 1805
Logistic regression 884 890 14 17 1805
Neural network 882 894 10 19 1805
True label True label
True label
True label True label
Journal of Healthcare Engineering 11
Table 6: Summary of results of different performance measures of applied machine learning techniques.
ML algorithm Accuracy Precision Recall F1 score AUC score RMSE
Support vector machine 0.97673 0.98 0.97 0.97 0.9766 0.1525
Decision tree 0.98560 0.99 0.99 0.99 0.9856 0.1200
Naı̈ve Bayes 0.98227 0.98 0.98 0.98 0.9822 0.1331
Logistic regression 0.98473 0.98 0.97 0.98 0.9821 0.1310
Neural network 0.98393 0.98 0.98 0.98 0.9839 0.1267
Accuracy: accuracy is one of the most important per- false-positive rate.+e area under the ROC is known as ROC
formance evaluation measures used to calculate the per- curve and is used to measure the classifier’s efficiency. +e
formance of any machine learning algorithm. It is computed better classifier is the one whose area is closer to 1, and
as the total number of correctly classified instances divided Figure 5: ROC curves of applied machine learning algo-
by all instances’ summation. Mathematically, it is denoted as rithms (a, b, c, d, e) shows the ROC curves of different
follows: classifiers. +e area under the curve (AUC) is another
TP + TN measure to compute the performance of the machine
Accuracy � . (1) learning technique to distinguish between the labels, and
TP + TN + FP + FN mathematically, it is computed as follows:
Precision: the efficiency of the supervised machine TP
learning algorithm is measured through several performance True Positive Rate � ,
measures; precision is among them. It is computed using the TP + FN (7)
correctly predicted positive values ratio to the total positive FP
values. Mathematically, it can be represented as follows: False Positive Rate � .FP + TN
TP
Precision � . (2) AUC-ROC gives us the complete representation of
TP + FP confusion matrices at different points in the graph. A
Recall: it is another performance measure for calculating confusion matrix is given at particular point, but AUC gives
the efficiency of a supervised machine learning algorithm. It us graphical representation of confusion matrices at various
is the ratio of correctly predicted positive values to all values threshold points.+e drawn line should be close to the upper
of actual class. Mathematically, it is shown as follows: right corner, i.e., 1, the model’s good. In our case, almost all
the lines of applied machine learning algorithms are close to
TP
Recall � . (3) the upper right corner of the graph. So, the developed
TP + FN models have achieved good in terms of ROC-AUC curve.
F1 Score: the performance measure is used to calculate Figure 6, shows the performance evaluation of employed
the performance of a supervised machine learning algo- algorithms in terms of accuracy, precision, recall, F1 score,
rithm. It is computed with the help of two measures, i.e., root-mean-square error, and AUC of the different classifiers.
precision and recall. Mathematically given by the harmonic +e results in Table 4 and Figure 6 indicate that models built
mean of precision and recall, it is calculated as follows: using these five different machine learning algorithms on
our dataset had achieved above 97% accuracy. +e decision
Precision∗Recall
F1 − Score � 2∗ , (4) tree has achieved 98.5%, SVM had shown 97%, and the rest
Precision + Recall have shown 98% accuracy, and other values are also good for
where all algorithms.+e results have shown that this model will be
effective in predicting the COVID-19 suspects in early
TP stages.
Precision � ,
TP + FP +e graphs for different performance measures such as
(5) accuracy, precision, recall, F1 score, RMSE, and AUC score
TP
Recall � . are shown above. One graph is corresponding to one per-
TP + FN formancemeasure such as the accuracy of all appliedmachine
Root Mean Square Error: RMSE is another performance learning algorithms to clearly visualize the output. Similarly,
measure used to calculate the performance of a supervised other graphs have been drawn to visualize the other per-
machine learning algorithm. Mathematically, it is computed formance measures. In terms of accuracy, the decision tree
as follows: has achieved the highest, SVM has achieved the lowest, andthe rest have shown equal. +e precision, recall, and F1 score
􏽲�����������������
FP + FN of proposed algorithms are the highest of decision tree, and
RMSE � . (6) the rest are at the same place. In case of root-mean-square
TP + TN + FP + FN error, the decision tree has the lowest followed by neural
Receiver Operating Characteristic: this curve is another network, logistic regression, and näıve Bayes, and the support
performance measurement criterion for measuring the ef- vector machine has the highest root-mean-square error value.
ficiency of machine learning classification algorithm. ROC is +e SVM has achieved low in terms of area under the curve
drawn by representing the true-positive rate against the score, and the rest are at the same place.
12 Journal of Healthcare Engineering
1.0 1.0
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
False Positive Rate False Positive Rate
SVM DT
(a) (b)
1.0 1.0
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
False Positive Rate False Positive Rate
MNB LR
(c) (d)
1.0
0.8
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
False Positive Rate
NN
(e)
Figure 5: ROC curves of applied machine learning algorithms. (a) ROC curve of SVM. (b). ROC curve of decision tree. (c). ROC curve of
näıve Bayes. (d). ROC curve of logistic regression. (e). ROC curve of neural network.
6. Comparative Analysis proposed hybrid model ResNet-SegNet has achieved the
highest accuracy of 99% [53].+e authors have proposed the
+e proposed work is novel, and the dataset used during the robust and stable inter-variability of CT lung image seg-
experimentation is primarily collected from patients. +e mentation of COVID-19 to avoid bias. +e study uses two
data collected are symptomatic, which is to be used to train ground truth (GT) annotations of chest images.+e three AI
the machine learning model to detect and predict the models trained are PSPNet, VGG-SegNet, and ResNet-
COVID-19 suspect in early stages to eliminate the mortality SegNet on GT annotations. +e ResNet-SegNet has per-
and spread of the infection. +e work is compared with the formed well in comparison with the other two. Reference
three different papers based on common parameters. +e [54] is a systematic review of AI technologies with respect to
papers [24, 53, 54] used for comparison are the best papers ARDS-COVID-19. +e dataset of CT images of lungs has
that can be taken as the benchmark in the field of deep been studied to understand the risk of bias (RoB) in a
learning. +e authors have used computed tomography (CT nonrandomized AI trial for handling ARDS using novel
scan) image set as a dataset. Reference [24] has used hybrid AtheroPoint-AI-Bias (AP(ai)Bias). Reference [55] has taken
deep learning AI models for lung image segmentation such the dataset of positive patients only and has trained the
as SegNet, VGG-SegNet, ResNet-SegNet, and NIH. +e machine learning model. In this study, SVM and decision
True Positive Rate True Positive Rate
True Positive Rate
True Positive Rate True Positive Rate
Journal of Healthcare Engineering 13
ACCURACY PRECISION
100 1
99.5 0.995
99 0.99
98.5 0.985
98 0.98
97.5 0.975
97 0.97
96.5 0.965
96 0.96
95.5 0.955
95 0.95
Support Vector Decision Naïve Logistic Neural Support Vector Decision Naïve Logistic Neural
Machine Tree Bayes Regression Network Machine Tree Bayes Regression Network
(a) (b)
RECALL F1-SCORE
1 1
0.995 0.995
0.99 0.99
0.985 0.985
0.98 0.98
0.975 0.975
0.97 0.97
0.965 0.965
0.96 0.96
0.955 0.955
0.95 0.95
Support Vector Decision Naïve Logistic Neural Support Vector Decision Naïve Logistic Neural
Machine Tree Bayes Regression Network Machine Tree Bayes Regression Network
(c) (d)
ROOT MEAN SEQUARED ERROR AREA UNDER CURVE SCORE
16 1
0.995
15 0.99
14 0.985
0.98
13 0.975
0.97
12 0.965
11 0.96
0.955
10 0.95
Support Vector Decision Naïve Logistic Neural Support Vector Decision Naïve Logistic Neural
Machine Tree Bayes Regression Network Machine Tree Bayes Regression Network
(e) (f )
Figure 6: Performance evaluation of employed algorithms.
Table 7: Comparative analysis.
Authors Technique Dataset AUC Accuracy
[24] Suri et al. AI, deep learning CT images ✓ ✓
[56] Otoom et al. Machine learning Only positive data ✓ ✓
[54] Suri et al. AI, deep learning CT images NA NA
[55] Zoabi et al. Machine learning Less number of attributes ✓ NA
[53] Suri et al. AI, deep learning CT images ✓ NA
Mir et al. (Proposed) Machine learning Primary symptomatic data ✓ ✓
table have achieved an accuracy of 93.0% and the rest are 7. Conclusion
below them. In terms of ROC area, the decision table has got
the highest of 95.5% and the rest have achieved below it. +e article proposes the framework to identify and predict
Reference [56] has been used in their work, but with less the COVID-19 suspect early to eliminate the mortality and
number of attributes. Table 7: comparative analysis, of the spread of infection. +e proposed framework collects the
proposed work with the above works already done in these data from sensors and IoT devices and employs machine
papers, is detailed as follows. learning to detect and predict COVID-19 suspect. +e
14 Journal of Healthcare Engineering
framework comprises logically connected four components: [6] M. Poongodi, M. Malviya, M. Hamdi, H. T. Rauf, S. Kadry,
data collection layer, data analytic center, diagnostic system, and O. +innukool, “+e recent technologies to curb the
and cloud system. +e framework is tested using machine second-wave of COVID-19 pandemic,” IEEE Access, vol. 9,
learning algorithms on a real dataset collected from SKIMS, pp. 97906–97928, 2021.
Srinagar. +e five proposed machine learning algorithms, [7] T. Ladd and O. Groth, .e internet of things, Econ, vol. 411,
support vector machine, decision tree, naı̈ve Bayes, logistic 8964 pages, UK, 2015.
regression, and neural network, have been used during our [8] A. S. R. Srinivasa Rao and J. A. Vazquez, “Identification of
study. +e experimental results have shown that all the ML COVID-19 can be quicker through artificial intelligenceframework using a mobile phone-based survey when cities
techniques have achieved above 97% accuracy. +e support and towns are under quarantine,” Infection Control & Hos-
vector machine has achieved 97.67%, the decision tree has pital Epidemiology, vol. 41, no. 7, pp. 826–830, 2020.
achieved 98.56%, and the rest have a round figure of 98%. [9] U. Iqbal and A. H. Mir, “Secure and scalable access control
+e decision tree has achieved good in other performance protocol for IoT environment,” Internet of .ings, vol. 12,
measures such as precision, recall, F1 score, root-mean- Article ID 100291, 2020.
square error, and area under the curve score. Keeping all the [10] A. Castiglione, M. Umer, S. Sadiq, M. S. Obaidat, and
performance measures under consideration, the decision P. Vijayakumar, “+e role of internet of things to control the
tree has performed well on our dataset among all proposed outbreak of COVID-19 pandemic,” IEEE Internet of .ings
techniques. +e proposed framework has the potential to Journal, vol. 8, no. 21, pp. 16072–16082, 2021.
eliminate and reduce the spread of infection through early [11] K. Kumar, N. Kumar, and R. Shah, “Role of IoT to avoid
detection and prediction system. +e data stored in cloud spreading of COVID-19,” International Journal of Intelligent
can easily be accessed by healthcare professionals to further Networks, vol. 1, pp. 32–35, 2020.
analyze it to get better insights and better understand the [12] U. Iqbal and A. H. Mir, “Efficient and dynamic access control
nature of disease. In future, our focus will be to propose mechanism for secure data acquisition in IoT environment,”
ensemble approaches such as random forest and various International Journal of Computing and Digital Systems,vol. 10, no. 1, pp. 9–28, 2021.
gradient boosting algorithms to train our algorithms. +e [13] A. Sufian, A. Ghosh, A. Safaa, and F. Smarandache, Since
dataset used in the above work is not so big that it will be January 2020 Elsevier Has Created a COVID-19 Resource
good to use ensemble learning or other methods. Further- centre with Free Information in English and Mandarin on the
more, deep learning techniques will also be experimented for Novel Coronavirus COVID- 19. .e COVID-19 Resource
enhancing the performance measures of the model. centre Is Hosted, Elsevier Connect, the Company’ s Public
News and Information, Amsterdam, Netherlands, 2020.
Data Availability [14] M. U. Ashraf, A. Hannan, S. M. Cheema, Z. Ali, K. M. Jambi,
and A. Alofi, “Detection and tracking contagion using IoT-
+e data will be made available on request from the cor- edge technologies: confronting COVID-19 pandemic,” in
responding author. Proceedings of the 2020 International Conference on Electrical,
Communication, and Computer Engineering (ICECCE),
Conflicts of Interest Istanbul, Turkey, June 2020.[15] U. Iqbal and A. Hussain Mir, “Secure and practical access
+e authors declare that there are no conflicts of interest control mechanism for WSN with node privacy,” Journal of
regarding the publication of this study. King Saud University - Computer and Information Sciences,
2020.
[16] A. A. Hussain, O. Bouachir, F. Al-Turjman, and M. Aloqaily,
References “Notice of retraction: AI techniques for COVID-19,” IEEE
[1] D. Cucinotta and M. Vanelli, “WHO declares COVID-19 a Access, vol. 8, pp. 128776–128795, 2020.
pandemic,” Acta Bio-Medica: Atenei Parmensis, vol. 91, no. 1, [17] J. Bhola, S. Soni, and G. K. Cheema, “Recent trends for se-
pp. 157–160, 2020. curity applications in wireless sensor networks – a technical
[2] S. Zhang and J. Lian, “Clinical characteristics of different review,” in Proceedings of the 2019 6th International Con-
subtypes and risk factors for the severity of illness in patients ference on Computing for Sustainable Global Development
with COVID-19 in Zhejiang, China,” Infectious Diseases of (INDIACom), pp. 707–712, New Delhi, India, March 2019.
Poverty, vol. 9, no. 1, pp. 1–10, 2020. [18] X. Huang, V. Jagota, E. Espinoza-Muñoz, and J. Flores-
[3] N. C. Peeri, N. Shrestha, M. S. Rahman et al., “+e SARS, Albornoz, “Tourist hot spots prediction model based on
MERS and novel coronavirus (COVID-19) epidemics, the optimized neural network algorithm,” International Journal of
newest and biggest global health threats: what lessons have we System Assurance Engineering and Management, 2021.
learned?” International Journal of Epidemiology, vol. 49, no. 3, [19] S. Tuli, S. Tuli, R. Tuli, and S. Singh, Since January 2020
pp. 717–726, 2020. Elsevier Has Created a COVID-19 Resource centre with Free
[4] W. Guan and Z. Ni, “Clinical characteristics of coronavirus Information in English and Mandarin on the Novel Corona-
disease 2019 in China,” New England Journal of Medicine, virus COVID- 19. .e COVID-19 Resource centre Is Hosted,
vol. 382, no. 18, pp. 1708–1720, 2020. Elsevier Connect, the Company’ s Public News and Infor-
[5] R. Bharti, A. Khamparia, M. Shabaz, G. Dhiman, S. Pande, mation, Amsterdam, Netherlands, 2020.
and P. Singh, “Prediction of heart disease using a combination [20] S. A.-F. Sayed, A. M. Elkorany, and S. Sayed Mohammad,
of machine learning and deep learning,” Computational In- “Applying different machine learning techniques for pre-
telligence and Neuroscience, vol. 2021, Article ID 8387680, diction of COVID-19 severity,” IEEE Access, vol. 9,
11 pages, 2021. pp. 135697–135707, 2021.
Journal of Healthcare Engineering 15
[21] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari, and [37] M. Javaid and I. H. Khan, “Internet of +ings (IoT) enabled
M. Ayyash, “Internet of things: a survey on enabling tech- healthcare helps to take the challenges of COVID-19 Pan-
nologies, protocols, and applications,” IEEE Communications demic,” Journal of Oral Biology and Craniofacial Research,
Surveys & Tutorials, vol. 17, no. 4, pp. 2347–2376, 2015. vol. 11, no. 2, pp. 209–214, 2021.
[22] Y. Dong and Y.-D. Yao, “IoTplatform for covid-19 prevention [38] R. Sujatha, B. Venkata Siva Krishna, J. Moy Chatterjee et al.,
and control: a survey,” IEEE Access, vol. 9, pp. 49929–49941, “Prediction of suitable candidates for covid-19 vaccination,”
2021. Intelligent Automation & Soft Computing, vol. 32, no. 1,
[23] P. N. Mahalle and S. S. Sonawane, “Internet of things in pp. 525–541, 2022.
healthcare,” Springer Briefs in Applied Sciences and Technol- [39] R. P. Singh, M. Javaid, A. Haleem, and R. Suman, “Internet of
ogy, vol. 22, pp. 13–25, 2021. things (IoT) applications to fight against COVID-19 pan-
[24] J. S. Suri, S. Agarwal, R. Pathak et al., “Covlias 1.0: lung demic,” Diabetes & Metabolic Syndrome: Clinical Research
segmentation in COVID-19 computed tomography scans Reviews, vol. 14, no. 4, pp. 521–524, 2020.
using hybrid deep learning artificial intelligence models,” [40] P. Verma and S. K. Sood, “Cloud-centric IoT based disease
Diagnostics, vol. 11, no. 8, p. 1405, 2021. diagnosis healthcare framework,” Journal of Parallel and
[25] R. K. Garg, J. Bhola, and S. K. Soni, “Healthcare monitoring of Distributed Computing, vol. 116, pp. 27–38, 2018.
mountaineers by low power Wireless Sensor Networks,” [41] B. Wang, X. Yao, Y. Jiang, C. Sun, andM. Shabaz, “Design of a
Informatics in Medicine Unlocked, vol. 27, Article ID 100775, real-time monitoring system for smoke and dust in thermal
2021. power plants based on improved genetic algorithm,” Journal
[26] R. Krishnamoorthi, S. Joshi, H. Z. Almarzouki et al., “A novel of Healthcare Engineering, vol. 2021, Article ID 7212567,
diabetes healthcare disease prediction framework using ma- 10 pages, 2021.
chine learning techniques,” Journal of Healthcare Engineering, [42] F. Abdali-Mohammadi, M. N. Meqdad, and S. Kadry, “De-
vol. 2022, Article ID 1684017, 10 pages, 2022. velopment of an IoT-based and cloud-based disease predic-
[27] J. M. Chatterjee, R. Sujatha, “role of artificial intelligence in tion and diagnosis system for healthcare using machine
COVID-19 prediction based on statistical methods,” in Ap- learning algorithms,” IAES International Journal of Artificial
plications of Artificial Intelligence in COVID-19, S. Nandan Intelligence, vol. 9, no. 4, pp. 766–771, 2020.
Mohanty, S. K. Saxena, S. Satpathy, and J. M. Chatterjee, Eds., [43] T. T. Nguyen, Q. V. H. Nguyen, D. T. Nguyen, E. B. Hsu,
Med. Vir, 2021. S. Yang, and P. Eklund, “Artificial intelligence in the battle
[28] N. Kumar, N. N. Das, D. Gupta, K. Gupta, and J. Bindra, against coronavirus (COVID-19): a survey and future re-
“Efficient Automated Disease Diagnosis Using Machine
Learning Models,” Journal of Healthcare Engineering, search directions,” pp. 1–14, 2020, http://arxiv.org/abs/2008.
vol. 2021, Article ID 9983652, 13 pages, 2021. 07343.
[29] M. H. Mir and D. Ravindran, “Role of IoT in smart city [44] Y. Siriwardhana, C. De Alwis, G. Gur, M. Ylianttila, and
applications: a review,” Int. J. Adv. Res. Comput. Eng. M. Liyanage, “+e fight against the COVID-19 pandemic with
Technol.vol. 6, no. 7, pp. 1099–1104, 2017. 5G technologies,” IEEE Engineering Management Review,
[30] R. Sujath, J. M. Chatterjee, and A. E. Hassanien, “A machine vol. 48, no. 3, pp. 72–84, 2020.
learning forecasting model for COVID-19 pandemic in In- [45] N. Mukati, N. Namdev, R. Dilip, N. Hemalatha, V. Dhiman,
dia,” Stochastic Environmental Research and Risk Assessment, and B. Sahu, “Healthcare assistance to covid-19 patient using
vol. 34, no. 7, pp. 959–972, 2020. internet of things (iot) enabled technologies,” Mater Today
[31] G. Aceto, V. Persico, and A. Pescapé, “Industry 4.0 and health: Proc, 2020.
internet of things, big data, and cloud computing for [46] C. Chakraborty and A. Abougreen, “Intelligent internet of
healthcare 4.0,” Journal of Industrial Information Integration, things and advanced machine learning techniques for covid-
vol. 18, pp. 100129–100214, 2020. 19,” EAI Endorsed Transactions on Pervasive Health and
[32] S. M. Riazul Islam, D. Daehan Kwak, M. Humaun Kabir, Technology, vol. 7, no. 26, pp. 168505–168514, 2021.
M. Hossain, and K. S. Kyung-Sup Kwak, “+e internet of [47] E. Elbasi, S. Mathew, A. E. Topcu, and W. Abdelbaki, “A
things for health care: a comprehensive survey,” IEEE Access, survey on machine learning and internet of things for
vol. 3, pp. 678–708, 2015. COVID-19,” in Proceedings of the 2021 IEEE World AI IoT
[33] C. Iwendi, A. K. Bashir, A. Peshkar et al., “COVID-19 patient Congress (AIIoT), pp. 115–120, Charlotte, NC, USA, January
health prediction using boosted random forest algorithm,” 2021.
Frontiers in Public Health, vol. 8, pp. 1–9, 2020. [48] S. Vyas and D. Bhargava, “Big data analytics and cognitive
[34] P. Verma, S. K. Sood, and S. Kalra, “Cloud-centric IoT based computing in smart health systems,” in Smart Health Sys-
student healthcare monitoring framework,” Journal of Am- temsSpringer, Singapore, 2021.
bient Intelligence and Humanized Computing, vol. 9, no. 5, [49] S. Dhamodharavadhani, R. Rathipriya, and J. M. Chatterjee,
pp. 1293–1309, 2018. “COVID-19 mortality rate prediction for India using statis-
[35] S. Nizetic, P. Solic, D. L.-I. Gonzalez-de-Artaza, and tical neural network models,” Frontiers in Public Health,
L. Patrono, Since January 2020 Elsevier Has Created a COVID- vol. 8, pp. 1–12, 2020.
19 Resource centre with Free Information in English and [50] A. Aljumah, “Assessment of machine learning techniques in
Mandarin on the Novel Coronavirus COVID- 19. .e COVID- IoT-based architecture for the monitoring and prediction of
19 Resource centre Is Hosted, Elsevier Connect, the Company’ COVID-19,” Electronics, vol. 10, no. 15, pp. 1834–15, 2021.
s Public News and Information, 2020. [51] A. P. Ramallo-González, A. González-Vidal, and
[36] S. R. Anan, A. Hossain, Z. Milky, M. M. Khan, M. Masud, and A. F. Skarmeta, “Ciotvid: towards an open iot-platform for
S. Aljahdali, “Research and development of an iot-based re- infective pandemic diseases such as COVID-19,” Sensors,
mote asthma patient monitoring system,” Journal of vol. 21, no. 2, pp. 484–513, 2021.
Healthcare Engineering, vol. 2021, Article ID 2192913, [52] M. Ndiaye, S. S. Oyewobi, A. M. Abu-Mahfouz, G. P. Hancke,
21 pages, 2021. A. M. Kurien, and K. Djouani, “IoT in the wake of Covid-19: a
16 Journal of Healthcare Engineering
survey on contributions, challenges and evolution,” IEEE
Access, vol. 8, pp. 186821–186839, 2020.
[53] J. Suri, S. Agarwal, P. Elavarthi et al., “Inter-variability study of
covlias 1.0: hybrid deep learning models for covid-19 lung
segmentation in computed tomography,” Diagnostics, vol. 11,
no. 11, pp. 2025–11, 2021.
[54] J. S. Suri, S. Agarwal, S. Gupta et al., “Systematic review of
artificial intelligence in acute respiratory distress syndrome
for COVID-19 lung patients: a biomedical imaging per-
spective,” IEEE Journal of Biomedical and Health Informatics,
vol. 25, no. 11, pp. 4128–4139, 2021.
[55] Y. Zoabi, S. Deri-Rozov, and N. Shomron, “Machine learning-
based prediction of COVID-19 diagnosis based on symp-
toms,” Npj Digital Medicine, vol. 4, no. 1, pp. 1–5, 2021.
[56] M. Otoom, N. Otoum, M. A. Alzubaidi, and Y. Etoom, “An
IoT-Based Framework for Early Identification and Moni-
toring of COVID-19 Cases,” Biomedical Signal Processing and
Control, vol. 62, 2020.