Proceedings of the International Conference on Automation, Computing and Renewable Systems (ICACRS 2022)

IEEE Xplore Part Number: CFP22CB5-ART: ISBN: 978-1-6654-6084-2

Heart Disease Prediction and Classification using

Machine Learning and Transfer Learning Model
2022 International Conference on Automation, Computing and Renewable Systems (ICACRS) | 978-1-6654-6084-2/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICACRS55517.2022.10029279

R Sivaprasad Dr. M.Hema Dr Bharati N Sunil D M Dr. Vaishali Mochammad

Department of Assistant Professor Ganar Department of ECE Mehta Fahlevi
Electrical and Department of ECE Department of Assistant Professor Professor Department of
Electronics JNTU-GV College Anesthesia. REVA university Department of Management,
Engineering of Engineering, Associate India Computer Science BINUS Online
Associate Professor India professor. [email protected] and Engineering Learning,
Sri Sairam hema.asrith@gmail ESIC medical .in Panipat Institute of Bina Nusantara
Engineering .com College Engineering and University,
College India Technology, Indonesia
India drbharativm74@g India mochammad.fahlev
sivaprasad.eee@sai mail.com wadhwavaishali@g [email protected]
ram.edu.in mail.com

Abstract— Heart disease is a dangerous condition that can the IT solutions makes it possible to access the data from
lead to a fatal condition due to cardiac arrest. Recent studies different locations to produce the intelligence [2]. The
have revealed various facts for analyzing cardiac data by sensing, dimension of data also gets changed to another form named big
monitoring, and learning data in IoT to predict early diagnosis data, which represent the huge in size but also in schema [3].
and treatment. Through machine learning based feature analysis, The organizations maintain big data in various network
accurate disease detection has been implemented. However, the locations, which can be accessed to produce intelligence
dominant methods do not accurately predict the result since the towards decision making or anything. To provide access to the
incorrect features contain non-related support values to select the big data there are number of protocols has been discussed
features to perform training validation and produce prediction
inaccuracy. To overcome this limitation, a Machine Learning and earlier by various researchers
Transfer Learning Model (TLM) is proposed to perform heart The heart plays an important role in life. Diagnosis and
disease prediction. Initially, pre-processing has been carried prediction of heart related diseases should be very accurate [4],
out to reduce dimension, and the scaling factor was also used to complete and accurate because even the slightest mistake can
calculate the margin rate. To increasing the prediction accuracy lead to fatigue and human death. There are many deaths related
Disease Prone Impact Rate (DPIR) intends to find the support to heart and the number is increasing day by day. To solve this
values. To select the labeled features, Relative Feature Margin problem, a prognosis system for disease awareness is needed.
S election (RFMS ) is used to select and train the model by
Multilayer perception neural network (MLPNN). This classifier Machine learning is a field of artificial intelligence (AI) that
selects the margin weights to predict the heart disease risk level provides first-class support in predicting all types of events
based on the class. This predicts higher impact of cardiac trained from natural phenomena [5]. In this paper, we calculate
deficiency rate by attaining the relevant features based deep the accuracy of machine learning methods for predicting heart
feature data learning model, which produce higher precision rate disease. This algorithm uses the UCI repository database for
to increase the prediction accuracy than other methods. training and testing k-near neighbors, end tree, linear
regression, support vector machine (SVM) [6]. Anaconda
Keywords— heart disease prediction, feature selection and
(jupytor) notebook is the perfect tool for executing Python
classification, machine learning and transfer learning, neural
programming. This tool has a variety of libraries and title files
network.
that make your work more accurate and precise.

I. INT RODUCT ION Hygiene is an integral part of human life. Because of the
large amount of mental health data available in the healthcare
The growth of information technology has been adapted for industry, machine learning models are used to make effective
several problems. The medical industries use the growth of IT decisions in predicting heart disease [7]. By us ing machine
in several ways. The health care data has been maintained in learning techniques, you can reliably classify healthy and
different locations of any organizational units and it has been unhealthy individuals. In this study, a framework is developed
accessed through modern Internet. Through the Heart data for understanding the principles of predicting a patient's risk
collected information [1], data analysis is the important part in profile using clinical data parameters [8].
the disease prediction and recommendation has been performed
by several units which use the data present in different data The proposed model was developed using the machine
servers located geographically in different locations. However, learning and transfer learning statistic model. Excess fit and fit

978-1-6654-6084-2/22/$31.00 ©2022 IEEE 595

Authorized licensed use limited to: Yale University Library. Downloaded on October 21,2024 at 18:29:01 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Automation, Computing and Renewable Systems (ICACRS 2022)
IEEE Xplore Part Number: CFP22CB5-ART: ISBN: 978-1-6654-6084-2

related issues are resolved. This model shows excellent results systems to predict the cardiac disease for early diagnose
in both experimental and training data [9]. ANN were used to treatment. There exists several techniques in supporting such
analyze the performance of models that accurately predict the analytics and decisive support systems; still they suffer to
presence or absence of heart disease. achieve higher performance in disease prediction and
generating the analysis. Towards supporting the health care
II. RELAT ED WORK solutions various decision models are presented earlier but to
produce higher support to the decision-making process, it is
The heart disease prediction using Sequential Backward necessary to utilize huge amount of data in terms of big data.
Selection Detection of Heart Disease (HD) models using For a hospital unit, maintaining such huge data becomes a
Machine Learning (ML) will be very useful in the early stages tedious but still the big data can be accessed towards analyzing
[10]. If it is detected at an early stage of the disease [11], HD the bio signals obtained from human body towards detection
models will be useful for treatment and recovery. HD symbolic and prediction of various diseas es.
machine learning technique has been developed to help
doctors.
RF Conjugant feature Multil
In [12] presented a reinforcement learning technique Inpu MS
towards routing in WSN which finds the path to the destination scale factor (CFSF) ayer
t - percep
and selects a route according to hop count, energy and distance. logs ML
[13] Presented a CHD detection approach over machine tion
PN neural
learning which uses ensembles towards disease prediction [14].
Similarly, a hybrid approach is presented to predict cardiac N Disease prone netwo
diseases with machine learning. The method combines several rk
impact rate (DPIR)
techniques towards disease prediction [15]. \
The effective heart disease prediction using hybrid model \
Heart disease is one of the most important causes of death in
the world today. Cardiovascular disease is a major challenge in Relative feature
the field of prediction and analysis of clinical data [16]. The Data
Novel Feature Reduction using machine learning and data logs margin selection
mining algorithms help in risk prediction. Error rate (RFMS)
mechanism of feature selection algorithm and analysis methods
for recommending new feature will reduce the rating process
which combines subgroups [17]. A metric profound influence Optimized prediction
on the choice of different feature selection algorithm that Patient
Monitoring class
evaluates algorithms vary in the middle.
The Monitoring and Heart Disease Prediction System
Using Deep Learning Modified Neural Network (DL-MNN) Figure 1 Proposed architecture diagram RFMS- MLPNN
[16] with the increasing popularity of smart wearable gadgets
offers internet (population) has been shown to present a The heart disease factors are labeled as features in dataset
number of solutions Internet of Things. Unfortunately, the to process the data. The transfer leaning base on artificial
chance of survival is lower for people suffering from sudd en neural network is optimized with multilayer perception be used
cardiac attacks. to analyze the data. When they collect data from medical
dataset which contains heart disease features of the patients’
The use of intelligent Machine learning system based tested values which is directly observed. The preprocessing
random forest algorithm helps in improving heart disease was carried to verify the presence of all the features, filling,
diagnosis. The main cause of Heart Failure (HF) is due to removing, indexing was carried out. This research is about the
narrowing or narrowing of the coronary arteries [17]. The development of heart disease prediction towards accessing big
coronary artery distributes blood to the heat. The stacking and data and how they can be analyzed towards health care.
Non-invasive detection of coronary artery disease, Coronary
Heart Disease (CHD), layering model and invasive technique
3.1 Conjugant feature scale factor (CFSF)
for diagnosis are accurate [18]. However, the invasive
procedure is not suitable for detecting coronary artery disease This stage choose the feature limits from the preprocessed
during an annual physical investigation [19]. Predict dataset. These creates marginal scaling values between higher
cardiovascular events in hypertensive patients to explain the and lower class based on defect ranges from medical domain.
prognosis of patients with hypertension, and is considered to be Based on the margins the features get demilitarized to group
critical to the development of cardiovascular disease the importance of feature labels. This selects the relatively high
prevention. margins based on average mean rate estimation to attain the
clusters as upper and lower boundary values. This reduces
feature dimension to conjugate the disease weightage values to
III. 3. PROPOSED SYST EM
predict.
This research importance is to heart disease prediction
based on machine and transfer learning and health care support

978-1-6654-6084-2/22/$31.00 ©2022 IEEE 596

Authorized licensed use limited to: Yale University Library. Downloaded on October 21,2024 at 18:29:01 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Automation, Computing and Renewable Systems (ICACRS 2022)
IEEE Xplore Part Number: CFP22CB5-ART: ISBN: 978-1-6654-6084-2

Algorithm steps
Begin For this type of support value consensus values is ,

Step 1: Import the weightage features

Step 2: For each instance
If i==1 then The selective prone subjectival;y attin
To calculate average mean rate target position Dpf

and

Feature selection and extraction usually select arbitrarily

Compute the ‘s’ points feature scaling margins is rate belongs to the dependencies of disease factor. Knowledge
referred to a margins high and low values. Denotes the of the search algorithm uses the learning process to guarantee
upper bounds, denotes lower bound. the extraction of features that reduce the dimensions of the data
and improve the classification results.
End if
3.3 Relative feature margin selection (RFMS)
To calculate the average confidence values from
target position medical margins This feature is selected based on the relevant subset
feature by measuring the critical seasonal dimension according
to relative Subset radial function (RSRF) without significantly
= ), n=1,2…,
reducing the accuracy of successful predictions using intensive
agro successive margins (IASM). This ability to capture
End for metabolic forecast data from the seasonal repository of cold
chain products for crops with the collection database
Step 3: For each i of l choose the recognizes that it is suitable for crop cultivation in a particular
Update the defect margins to indexing area.
Algorithm:

End for Step1: Initialize the process Mx-f successive feature impact
rate.
Step 4: Return risk margins of the heart disease scaling
factor. Step 2: Computing all margins from Mx-f;

Stop Step 3 For All Mx-fclass margins Mc; i=0, i++

The above algorithm selects the conjugant scaling factors, Step 4 Select the consecutive class Labels from
to reduce dimension of non-related feature margins, this Mx-f;
support for choosing relational features actively for disease Set max support Lm {s,p, d, ltr, humidity,
affection state. This much support for importance disease rainfall, temperature }
defect level.
End For
3.2 Disease prone impact rate (DPIR) If Lm  Mx-f (class)
This stage the features get decisional into disease affection
Select feature to relative margin (Rm)
rate. The prone to find the risk level relatively taken from
cardiac principle. This selects the prone attain levels cardiac End If
deficiency rate weight is centralized to make comparison and
decision to attain the weight, such as finding the exact Step 5 Construct relative Subset radial function (Srf Mx-
f (RM))
characteristics of a particular transformation ( ) or
developing an evaluative observing importance dense level For each Process the Low margin Successive mean rate
from ‘N’ levels. (Lms)
Create subset for each class Lms<= Mx-f (Rm)
Attain Lmin class (Lmin-c) and (Lmax-c)
Create the subset class L(Max-Min)
Subject prone factors, they a, b are relative margins with ‘y’
variants at constant levels. Split the partition Pi(L) cluster centroid (C)

978-1-6654-6084-2/22/$31.00 ©2022 IEEE 597

Authorized licensed use limited to: Yale University Library. Downloaded on October 21,2024 at 18:29:01 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Automation, Computing and Renewable Systems (ICACRS 2022)
IEEE Xplore Part Number: CFP22CB5-ART: ISBN: 978-1-6654-6084-2

End loop of (j). End

Step 6: Construct intensive agro successive margins (Pi(L)) The aspect values of the data sets are continuous and
the reciprocal information requires additional calculation for
Select the MX feature support weight W-max(s)
the reciprocal data, so instead of reciprocal information we use
Return W-max(s) two similarities for the similarity measure and two weight
margin for comparing the feature. Not all layers are linear.
The above algorithm selects the importance of the Activation function figures the internal result of th e input
feature to reduce the dimens ion by hybridizing the optimization vector and the heaviness of that unit constructs subset cluster
techniques. The selected features be further trained into neural based on pattern mapping. Only the synaptic weights must be
network. Based on the threshold margins the class get trained scholarly found to classify the result which produces high
with soft-max activation function and processed in deep performance.
learning neural network.

3.4 Multilayer perception neural network (MLPNN) IV. 4. RESULT AND DISCUSSION

This classifier identifies the risk of heart disease based on The proposed approach is implemented under python
the selected features trained into neural network. They testing framework using various parameters and performance is
construct transfer leering based neural network using multi- evaluated. This method measures efficiency in the prognosis of
kernel perceptron neural network with soft max decision a disease based on various features and their values. The testing
logical activation function. This sets the condition for each and training were carried through confusion matrix. The results
feature pattern disease influence rate depending on successive of the evaluation are analyzed in conjunction with the
patterns. performance of other approaches. The results will be displayed
in this section.
Step 1: Construct a set feature T to contain the T able 1 values and parameter processed
selected features. Initially T0 = ∅ (n features).
Parameter Value
Step 2: Initialize neural network with number of
neurons and features. Language, Tool Used Python,
Estimate feature support level
Dataset used CVD-cardiac dataset
Step 3: Analyze the disease as first selected feature
T1=max M threshold margin Number of patients 2000
Step 4: for i=2,.., m: Choose feature that maximizes
M Number of features 30/3000
/records
Add features to T then
Compute the number of feature occurrence attribute
pattern Pi. Table 1 shows the details used to evaluate the performance
generated in different ways. Accordingly, the method measures
//transfer terms, k times subset at feed forward performance by various limits. The consequences got are given
layers selection in detail in this section.
Step5: Compute combination of possible feature T able 2: Analysis on Mean precision and recall rate
patterns
Mean precision and recall rate in %
K- Specific features values.
Number of 500 1000 2000
Step6: for each pattern Pi from subset cluster nodes/Methods
factorScf
RF 65 70 77
Estimate the compound relation Crl = ø(Pi Scf))/m. SVM 70 74 83
Ø- Number of repeated pi contained in scf. HDPM 74 80 86
Train the Relation pattern (Rpn) = Crl/total count.
RFMS- 83 85 93
IfMaxTh support >Rpn then. MLPNN
Add Pi to extracted feature Crl.
SsAverage mean The performance of the routes generated by the
various methods was measured and shown in Table 2. Here, the
Return class by Patternpi(Rpn)Risk by class proposed RFMS- MLPNN algorithm produces higher routing
End if efficiency than other approaches.

End For.

978-1-6654-6084-2/22/$31.00 ©2022 IEEE 598

Authorized licensed use limited to: Yale University Library. Downloaded on October 21,2024 at 18:29:01 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Automation, Computing and Renewable Systems (ICACRS 2022)
IEEE Xplore Part Number: CFP22CB5-ART: ISBN: 978-1-6654-6084-2

Classification accuracy

RFMS- MLPNN

Methods
HDPM
200 records
SVM
100 records
RF 50 records
70 80 90 100
Accuracy

Figure 2: Mean precision and recall rate

Figure 3: Classification accuracy

Figure 3 shows the clustering performance generated

The routing performance introduced by different by different methods. The proposed RFMS- MLPNN approach
approaches at the nodes of different numbers in the network is has developed high clustering accuracy under a number of
measured and shown in Figure 2. The proposed RFMS- different diseases.
MLPNN approach has a higher routing efficiency than other T able 4: Analysis on Disease Prediction
methods at all levels.
T able 3: Classification accuracy
Disease Prediction Accuracy
Classification accuracy in %
Number of 500 1000 2000 Number of 50 100 200
Records Records
/methods /methods

RF 75 80 86 RF 64 70 76

SVM 81 85 90
HDPM 84 88 92 SVM 65 72 75

RFMS- 87 91 96
MLPNN HDPM 66 75 80

RFMS- 70 78 82
Table 3 shows the accuracy of clustering large data MLPNN
for disease prognosis. Here, the proposed RFMS- MLPNN
approach provides greater clustering accuracy than other
methods. The prognostic performance of the disease and its accuracy
are measured considering different disease classes. The results
obtained are shown in Table 4.4. The proposed RFMS-
MLPNN approach resulted in higher disease prognosis
accuracy than other methods.

978-1-6654-6084-2/22/$31.00 ©2022 IEEE 599

Authorized licensed use limited to: Yale University Library. Downloaded on October 21,2024 at 18:29:01 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Automation, Computing and Renewable Systems (ICACRS 2022)
IEEE Xplore Part Number: CFP22CB5-ART: ISBN: 978-1-6654-6084-2

False Ratio
50 records
50 100 records
45

False Ratio %
40 200 records
35
30
25
20
15
10
5
0
50 records

Figure 4: Analysis on disease prediction accuracy

The accuracy of the prognosis generated by the Figure 5: Analysis on False Classification Ratio+
various methods was measured and is shown in Figure 4. The
proposed Hybrid approach resulted in a higher disease The accuracy of disease prognosis made by different
prognosis than other approaches in each class. methods is measured and shown in Figure 5. The proposed
RFMS- MLPNN approach resulted in a higher disease
T able 5: Analysis on False Ratio
prognosis than other approaches in each class.
False Classification Ratio
V. CONCLUSION
Number of 500 1000 2000
Records To perform heart disease diagnosis, intelligent large data
analysis models reduce the dimension of data analysis. A smart
/methods
IoT sensor for heart patient data analysis in healthcare is based
RF 30 25 20 on hyper-spectral deep neural network. Efficient heart disease
prognosis helps to improve the effectiveness of data analysis
and prediction using ambiguous rules. It provides predictive
results to users, giving them the ability to generate predictive
SVM 34 27 22
attack rates for early detection methods. Feature evaluation-
based prediction is also very sophisticated based on the
Multilayer perception neural classification in the functional
HDPM 30 23 18 achieves high classification and predication rate. The proposed
RFMS- MLPNN produce higher performance by evaluating the
RFMS- 28 20 16 cardiac features than other methods. This recommends the risk
MLPNN level based on classes by predicting the feature class and
recommend risk-based disease diagnosis.

The percentage of misclassifications introduced in different REFERENCES

ways is measured and shown in Table 5. Here, the proposed [1] Singh and R. Kumar, "Heart Disease Prediction Using Machine
RFMS- MLPNN approach has lower error rates than other Learning Algorithms," 2020 International Conference on Electrical
methods. and Electronics Engineering (ICE3), 2020, pp. 452-457, doi:
10.1109/ICE348803.2020.9122958.
[2] P. Ramprakash, R. Sarumathi, R. Mowriya and S.
Nithyavishnupriya, "Heart Disease Prediction Using Deep Neural
Network," 2020 International Conference on Inventive Computation
T echnologies (ICICT ), 2020, pp. 666-670, doi:
10.1109/ICICT 48043.2020.9112443.
[3] C. Boukhatem, H. Y. Youssef and A. B. Nassif, "Heart Disease
Prediction Using Machine Learning," 2022 Advances in Science
and Engineering Technology International Conferences (ASET ),
2022, pp. 1-6, doi: 10.1109/ASET 53988.2022.9734880.
[4] J. Yu, S. Park, S. -H. Kwon, K. -H. Cho and H. Lee, "AI-Based
Stroke Disease Prediction System Using ECG and PPG Bio -
Signals," in IEEE Access, vol. 10, pp. 43623-43638, 2022, doi:
10.1109/ACCESS.2022.3169284.

978-1-6654-6084-2/22/$31.00 ©2022 IEEE 600

Authorized licensed use limited to: Yale University Library. Downloaded on October 21,2024 at 18:29:01 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Automation, Computing and Renewable Systems (ICACRS 2022)
IEEE Xplore Part Number: CFP22CB5-ART: ISBN: 978-1-6654-6084-2

[5] S. Huang, P. Hung, C. Hong and H. Wang, "A New Image Blood Conference on Computing for Sustainable Global Development
Pressure Sensor Based on PPG, RRT , BPT T , and Harmonic (INDIACom), 2015, pp. 704-706.
Balancing," in IEEE Sensors Journal, vol. 14, no. 10, pp. 3685 -
3692, Oct. 2014, doi: 10.1109/JSEN.2014.2329676.
[6] S. S. Sarmah, "An Efficient IoT -Based Patient Monitoring and
Heart Disease Prediction System Using Deep Learning Modified
Neural Network," in IEEE Access, vol. 8, pp. 135784 -135797,
2020, doi: 10.1109/ACCESS.2020.3007561.
[7] N. T . Nguyen, N. V. Nguyen, M. H. T. Tran and B. T. Nguyen, "A
potential approach for emotion prediction using heart rate signals,"
2017 9th International Conference on Knowledge and Systems
Engineering (KSE), 2017, pp. 221-226, doi:
10.1109/KSE.2017.8119462.
[8] S. S. Sarmah, "An Efficient IoT -Based Patient Monitoring and
Heart Disease Prediction System Using Deep Learning Modified
Neural Network," in IEEE Access, vol. 8, pp. 135784-135797,
2020, doi: 10.1109/ACCESS.2020.3007561.
[9] V. Chacko and V. Bharati, "Data Validation and Sensor Life
Prediction Layer on Cloud for IoT ," 2017 IEEE International
Conference on Internet of T hings (iT hings) and IEEE Green
Computing and Communications (GreenCom) and IEEE Cyber,
Physical and Social Computing (CPSCom) and IEEE Smart Data
(SmartData), 2017, pp. 906-909, doi: 10.1109/iThings-GreenCom-
CPSCom-SmartData.2017.139.
[10] Ed-Daoudy and K. Maalmi, "Real-time machine learning for early
detection of heart disease using big data approach," 2019
International Conference on Wireless Technologies, Embedded and
Intelligent Systems (WIT S), 2019, pp. 1-5, doi:
10.1109/WIT S.2019.8723839.
[11] Q. He, A. Maag and A. Elchouemi, "Heart disease monit oring and
predicting by using machine learning based on IoT technology,"
2020 5th International Conference on Innovative T echnologies in
Intelligent Systems and Industrial Applications (CIT ISIA), 2020,
pp. 1-10, doi: 10.1109/CIT ISIA50690.2020.9371772.
[12] M. A. Khan and F. Algarni, "A Healthcare Monitoring System for
the Diagnosis of Heart Disease in the IoMT Cloud Environment
Using MSSO-ANFIS," in IEEE Access, vol. 8, pp. 122259-122269,
2020, doi: 10.1109/ACCESS.2020.3006424.
[13] J. P. Li, A. U. Haq, S. U. Din, J. Khan, A. Khan and A. Saboor,
"Heart Disease Identification Method Using Machine Learning
Classification in E-Healthcare," in IEEE Access, vol. 8, pp. 107562-
107582, 2020, doi: 10.1109/ACCESS.2020.3001149.
[14] S. A. Ali et al., "An Optimally Configured and Imp roved Deep
Belief Network (OCI-DBN) Approach for Heart Disease Prediction
Based on Ruzzo–Tompa and Stacked Genetic Algorithm," in IEEE
Access, vol. 8, pp. 65947-65958, 2020, doi:
10.1109/ACCESS.2020.2985646.
[15] Y. Pan, M. Fu, B. Cheng, X. T ao and J. Guo, "Enh anced Deep
Learning Assisted Convolutional Neural Network for Heart Disease
Prediction on the Internet of Medical T hings Platform," in IEEE
Access, vol. 8, pp. 189503-189512, 2020, doi:
10.1109/ACCESS.2020.3026214.
[16] N. L. Fitriyani, M. Syafrudin, G. Alfian and J. Rhee, "HDPM: An
Effective Heart Disease Prediction Model for a Clinical Decision
Support System," in IEEE Access, vol. 8, pp. 133034-133050, 2020,
doi: 10.1109/ACCESS.2020.3010511.
[17] S. Mohan, C. T hirumalai and G. Srivastava, "Effective Heart
Disease Prediction Using Hybrid Machine Learning Techniques," in
IEEE Access, vol. 7, pp. 81542-81554, 2019, doi:
10.1109/ACCESS.2019.2923707.
[18] X. Yuan, J. Chen, K. Zhang, Y. Wu and T . Yang, "A Stable AI-
Based Binary and Multiple Class Heart Disease Prediction Model
for IoMT," in IEEE Transactions on Industrial Informatics, vol. 18,
no. 3, pp. 2032-2040, March 2022, doi: 10.1109/TII.2021.3098306.
[19] Dewan and M. Sharma, "Prediction of heart disease using a hybrid
technique in data mining classification," 2015 2nd International

978-1-6654-6084-2/22/$31.00 ©2022 IEEE 601

Authorized licensed use limited to: Yale University Library. Downloaded on October 21,2024 at 18:29:01 UTC from IEEE Xplore. Restrictions apply.