International Conference on Innovative Data Communication Technologies and Application (ICIDCA-2023)

IEEE Xplore Part Number: CFP23CR5-ART; ISBN: 979-8-3503-9720-8

Machine Learning based Chronic Disease (Heart

Attack) Prediction
Dr. V Selvakumar
Advaith Achanta Nithin Sreeram
Assistant Professor , Bhavans Vivekananda College
Bhavans Vivekananda College of
2023 International Conference on Innovative Data Communication Technologies and Application (ICIDCA) | 979-8-3503-9720-8/23/$31.00 ©2023 IEEE | DOI: 10.1109/ICIDCA56705.2023.10099566

Bhavans Vivekananda College of of Science, Humanities &

Science, Humanities & Commerce Science, Humanities & Commerce
Hyderabad, India Commerce
Hyderabad, India Hyderabad, India
[email protected] [email protected]
[email protected]

Abstract— A heart attack is a life-threatening event that heart attack yearly, and approximately 30% of these
can be extremely difficult to predict. Early detection and incidents are recurrent events. These statistics highlight the
prompt treatment can significantly reduce mortality rates. critical need for accurate and timely identification of
According to the British Health Foundation (B.H.F.), 1 in individuals at risk of heart disease and heart attacks. The
14 people live globally with a heart or circulatory disease. development of machine learning-based predictive
Moreover, around 200 million people are estimated to live models for heart attack prediction can improve the
with coronary heart disease. Over the past decades, machine accuracy of risk assessment and enable timely preventive
learning (ML) techniques have been applied to measures, leading to better health outcomes and reduced
cardiovascular data to identify patterns and predict healthcare costs.
outcomes. In particular, ML algorithms have been used to
predict the risk of a heart attack. This paper aims to use II. LIT ERAT URE REVIEW
machine learning to predict the risk of a patient having a [1] The study highlights the potential of machine
heart attack. Data with various features such as age, gender, learning models in predicting heart attack risk based on
and cholesterol have been used to create a predictive model.
non-invasive physiological and demographic factors.
The model was trained and tested on various datasets to
[2]The study highlights the potential of machine learning
determine its accuracy and ability to predict the risk of a
models in predicting heart attack risk based on E.C.G.
heart attack. The findings of this study can be used to
develop more accurate methods for predicting the risk of a results and other medical history factors. [3]The study
heart attack and reduce the number of deaths caused by highlights the potential of machine learning models in
heart attacks. predicting heart attack risk based on demographic and
clinical factors. [4]The authors discuss the strengths and
Keywords— Heart attack, machine learning, prediction, limitations of different approaches and highlight the need
algorithm, chronic disease for more standardized data collection and feature selection
methods. [5] The authors discuss the challenges associated
with heart failure prediction and the potential of machine
learning models in improving the accuracy of risk
I. INT RODUCT ION
assessment. [6] The authors discuss the potential of
Heart disease is a significant global health concern that machine learning models in complementing traditional
remains as a leading cause of mortality. Early detection epidemiological methods for heart disease prediction. [7]
and accurate prediction of the risk of heart attacks can play The authors highlight the potential of machine learning
a crucial role in implementing preventive measures and algorithms to increase the precision of cardiac disease
reducing the associated health risks. Machine learning prediction. [8] The authors address the potential of
(ML) has emerged as a promising approach for developing machine learning models in enhancing the accuracy of
predictive models for various medical conditions, heart failure prediction. [9] The authors emphasize how
including heart disease. Over the years, researchers have machine learning models could increase accuracy and
explored the potential of ML algorithms in predicting the efficiency of heart disease diagnosis. [10] The authors
risk of heart attacks based on various physiological and review various models and provide an overview of their
demographic factors. This research paper aims to review accuracy and limitations. They also discuss the
heart attack prediction using machine learning techniques. importance of feature selection and data standardization.
Specifically, this study will discuss the features used in [11] The authors compare the accuracy of various
these models, the different algorithms employed, and the algorithms and discuss the importance of feature selection
performance metrics used to evaluate the effectiveness of and data preprocessing. [12] The authors discuss the
these models. Ultimately, this paper aims to provide potential of machine learning models in improving the
insights into the potential of ML-based heart attack accuracy of risk assessment and disease prediction. [13]
prediction, highlighting the challenges and opportunities The authors review several models' accuracy, discuss the
for further research in this area. value of feature selection, and optimize the model. [14]
Heart disease is a leading cause of death worldwide, The authors present an overview of various techniques
responsible for an estimated 17.9 million deaths annually, and discuss the significance of feature selection and
according to the World Health Organization (WHO). In the model optimization. [15] The authors compare the
United States alone, more than 600,000 people die yearly accuracy of various models and discuss the importance of
due to heart disease, accounting for one in every four
deaths. In addition, about 735,000 Americans experience a

979-8-3503-9720-8/23/$31.00 ©2023 IEEE

Authorized licensed use limited to: ST. JOSEPH ENGINEERING COLLEGE MANGALORE. Downloaded on March 11,2025 at 12:10:53 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Innovative Data Communication Technologies and Application (ICIDCA-2023)
IEEE Xplore Part Number: CFP23CR5-ART; ISBN: 979-8-3503-9720-8

feature selection and data standardization. [16] The authors b) Model Training: Several machine learning
discuss the importance of feature selection and provide an models will be trained using the preprocessed dataset,
overview of the accuracy of various models. [17] The including logistic regression, random forest, and neural
authors compare the accuracy of various models and networks. The hyperparameters of the models will be
discuss the importance of feature selection and data optimized using cross-validation and grid search
standardization. [18] Predictive modeling for heart disease techniques.
using machine learning techniques file.
c) Model evaluation: The trained models will be
evaluated using appropriate model evaluation metrics such
III. A IMS AND OBJECT IVES as accuracy, precision, recall, and F1-score. The models
will be compared to identify the most accurate and reliable
A. Aims model.
 To develop a machine learning model for
heart attack prediction that can accurately C. Algorithms
predict patients' risk of a heart attack.
 To compare the performance of different machine a) Logistic Regression: Logistic Regression is a
learning algorithms for heart attack prediction and machine-learning model for predicting heart attacks. It is a
identify the most accurate and reliable model. binary classification model that uses a sigmoid function to
output a probability between 0 and 1, which indicates the
 To identify the most important risk factors likelihood of a patient having a heart attack. The formula
associated with heart attack and develop a
for logistic regression is:
predictive model incorporating these unavoidable
factors.
Y = 1/(1+e^(-(b0+b1X1+b2X2+ ⋯.+bnXn)) )

B. Objectives Where Y is the predicted probability, b0 is the intercept, bi is

 To preprocess and clean the dataset, including the coefficients of the independent variables x1, x2, ..., xn.
handling missing data, normalizing variables, and
transforming data as necessary. b) K.N.N.: A well-liked machine learning technique
called Logistic K-Nearest Neighbors (K.N.N.) is
 To train and test several machine learning models, frequently employed for both classification and regression
including logistic regression, random forest, and problems. It is a non-parametric algorithm that makes no
neural networks, to predict patients' heart attack assumptions regarding the underlying data distribution. In
risk. the context of heart attack prediction, K.N.N. can be
 To compare the accuracy and performance of the utilized to predict whether a patient is at risk of having a
different models using appropriate evaluation heart attack based on their medical history and other
metrics such as accuracy and recall. relevant features. The algorithm works by locating the k
number of nearest d a t a points to the testing point in the
 To analyze the interpretability and explain the data space and classifying the test data point based on the
ability of the machine learning models for heart
majority class of its k-nearest neighbors.
attack prediction to aid clinical decision-making.
The formula for K.N.N. classification involves
computing the distance between the test data point and all
I V. M AT ERIALS AND M ETHODS other data points in the feature space, selecting the k-
A. Materials nearest data points to the test data point based on the
a) Datasets: The data for the study were retrieved computed distances, and classifying the test data point
from a trusted secondary source, and this consisted of based on the majority class of its k-nearest neighbors.
multiple features like age, sex, cp, trpbs, and fbs among
others. The entire dataset consisted of 303 rows and 14 c) Support Vector Machine (SVM): SVM is a
columns. powerful machine learning model that can handle complex
data sets with high dimensional spaces. It works by
b) Software: Python will be used for data separating the data into different classes using a hyper
preprocessing, model training, and evaluation. The Scikit- plane. In heart attack prediction, SVM can be used to
learn library will be used for implementing machine separate patients who are at risk of heart disease from those
learning algorithms, and other libraries, such as Pandas who are not. The formula for SVM is:
and NumPy, will be used for data manipulation and
analysis. w^T x+b=0

B. Methods Where w is the weight vector, x is the input vector, and

a) Data preprocessing: The dataset will be bis the bias.
preprocessed by handling missing data, normalizing
variables, and transforming data as necessary.

979-8-3503-9720-8/23/$31.00 ©2023 IEEE

Authorized licensed use limited to: ST. JOSEPH ENGINEERING COLLEGE MANGALORE. Downloaded on March 11,2025 at 12:10:53 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Innovative Data Communication Technologies and Application (ICIDCA-2023)
IEEE Xplore Part Number: CFP23CR5-ART; ISBN: 979-8-3503-9720-8

d) Decision Trees: Decision Trees are a simple yet

powerful machine learning model that can be used for heart
attack prediction. They work by dividing the data into smaller In the context of heart attack prediction, XGBoost can
subsets based on the values of the input variables. The be utilized to train multiple weak learners, such as decision
formula for decision trees involves a series of if-then trees, on different subsets of the training data and then
combine their outputs to make a final prediction. This
statements that define the rules for dividing the data.
approach can improve the accuracy and robustness of the
model.
e) Random Forest: Random Forest is an ensemble j) Neural Networks: Neural Networks are a set of
learning model that combines multiple decision trees to make algorithms that are designed to recognize patterns. They are
predictions. It is a popular model for heart attack prediction as used extensively in heart attack prediction because they can
it can handle nonlinear relationships between variables. The manage complex, high-dimensional data sets. The formula
formula for a random forest is a bit complex as it involves for a neural network is difficult, as it involves multiple
combining multiple decision trees . Still, the basic idea is to layers of interconnected neurons, but the basic idea is to
aggregate the predictions of several decision trees to produce use a set of input nodes to represent the features of the data
a more accurate result. and output nodes to represent the predictions.

f) Bagging: Bootstrap aggregating (Bagging) is a Y=f(w_2 f(w_1 x+b_1 )+ b_2)

machine learning ensemble method that combines multiple
models trained on different subsets of the training dataset to
improve the algorithm's overall performance. It is a variance- Where y is the predicted output, x is the input data,
reduction technique that reduces overfitting and enhances the w1, and w2 are the neural network weights, b1, and b2 are
stability of the model. the biases, and f is the activation function.
k) Recall/Sensitivity: For the most accurate model,
In the context of heart attack prediction, Bagging can be another metric called recall/sensitivity was checked,
utilized to train multiple decision tree models on different focusing on the false negative. Heart attack prediction is a
subsets of the training data and then combine their outputs sensitive issue, and Type-2 error, in this case, cannot be
to forecast future values. tolerated, as it is still fine if a person doesn't have a chance
g) AdaBoost: AdaBoost, short for Adaptive of being prone to a heart attack, but the model predicts
Boosting, is a machine learning ensemble method that that the person has because it isn't life-threatening for the
combines multiple weak classifiers to obtain a strong patient, but the other way around isn't tolerable. The recall
classifier. It is a boosting technique that sequentially trains value achieved was 95%
models by increasing the weights of misclassified
observations, aiming to improve the algorithm's overall
performance.

In the context of heart attack prediction, AdaBoost can

be utilized to train multiple weak classifiers, such as decision
stumps, on different subsets of the training data and then
combine their outputs to make a future prediction. V . A NALYSIS

h) GradientBoost: Gradient Boosting is a machine In this study, we aimed to develop a heart attack
learning ensemble method that combines multiple weak prediction model using machine learning techniques. Our
learners to obtain a strong learner. It is a boosting technique dataset included information from 303 patients, each with
that sequentially trains models by minimizing the loss 14 different features, such as age, gender, blood pressure,
function using gradient descent, aiming to improve the and cholesterol levels. We split the dataset into a training
algorithm's overall performance. set (80%) and a testing set (20%) to evaluate the
performance of our model.

In the context of heart attack prediction, Gradient

Boosting can be utilized to train multiple weak learners, First, we conducted exploratory data analysis (E.D.A.)
such as decision trees on different subsets of the training to gain insights into the distribution (Fig 1.) and
data and then combining their outputs to make a final relationships (Fig 2.) of the variables.
prediction.

i) XGBoost: XGBoost, short for eXtreme Gradient

Boosting, is a machine learning ensemble method that
combines multiple weak learners to obtain a strong
learner. It is an optimized implementation of the Gradient
Boosting algorithm that utilizes tree-based models and
advanced regularization techniques to improve the
accuracy and speed of the algorithm.

979-8-3503-9720-8/23/$31.00 ©2023 IEEE

Authorized licensed use limited to: ST. JOSEPH ENGINEERING COLLEGE MANGALORE. Downloaded on March 11,2025 at 12:10:53 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Innovative Data Communication Technologies and Application (ICIDCA-2023)
IEEE Xplore Part Number: CFP23CR5-ART; ISBN: 979-8-3503-9720-8

A. Figures Random Forest Classifier experimented with various

train-test split ratios (Table III.), of which 80–20 gave the
best result. Many models yielded their best results with the
80 – 20 split ratio; hence, it was chosen to be the standard
ratio uniformly for all the models.

T ABLE IV. BAGGING

No. of Estimators Accuracy
100 82.46
500 84.21
1000 84.21

Bagging Classifier experimented with various numbers

of estimators (Table IV.), of which 500 and 1000
estimators gave the best result.

T ABLE V. ADA BOOST

Fig. 1. Heat map of the features No. of Estimators Accuracy
100 85.96
500 82.45
1000 82.45

Adaptive Boosting Classifier experimented with various

numbers of estimators (Table V.) of which 100 estimators
gave the best result.

T ABLE VI. GRADIENTBOOST

No. of Estimators Accuracy
100 82.45
500 80.70
Fig. 2. Double bar plot of sex vs. count 1000 80.70

B. Experimentation Gradient Boosting Classifier experimented with various

numbers of estimators (Table VI.) of which 100 and 1000
Extensive experimentation was performed on all the
estimators gave the best result.
models to reach the suitable characteristics for the model.
The experiment for each model is displayed below. T ABLE VII. X.G.B.O.O.S.T .

T ABLE I. LOGISTIC REGRESSION No. of Estimators Accuracy

100 82.45
Train Test Split Accuracy 500 82.45
80 – 20 91.22 1000 82.45
75 – 25 87.32
70 - 30 89.32
Extreme Gradient Boosting Classifier was experimented
with various numbers of estimators (Table VII.) of which all
Logistic regression experimented with various train-test
three numbers of estimators gave similar results.
split ratios (Table I.), of which 80 – 20 gave the best result.
T ABLE VIII. NEURAL NETWORKS WITH ADAM OP TIMZER
T ABLE II. DECISION T REES
Train Test Split Accuracy Train- Archite Epoch Accuracy
Test cture s
80 – 20 82.45 Split
75 – 25 80.28
70 - 30 75.81
80-20 12-6-3-1 100 56.14

Decision Trees Classifier experimented with various 80-20 12-6-3-1 200 64.91
train-test split ratios (Table II.) of which 80 – 20 gave the
best result. 80-20 12-6-3-1 300 66.67

T ABLE III. RANDOM FOREST

80 – 20 12-6-3-1 400 63.16

Train Test Split Accuracy 80 -20 12-6-3-1 500 59.65

80 – 20 80.71
75 – 25 80.28
70 – 30 77.90

979-8-3503-9720-8/23/$31.00 ©2023 IEEE

Authorized licensed use limited to: ST. JOSEPH ENGINEERING COLLEGE MANGALORE. Downloaded on March 11,2025 at 12:10:53 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Innovative Data Communication Technologies and Application (ICIDCA-2023)
IEEE Xplore Part Number: CFP23CR5-ART; ISBN: 979-8-3503-9720-8

T ABLE IX. NEURAL NETWORKS WIITH SGD OP TIMIZER

It was observed that the age group of 50-60 had the most
Train Architectur Epoch Accurac
chance of being prone to a heart attack which, on further
- e s y descriptive analysis, was found to be rooted in the fact that
Te st most people retire during this age and our body gets so
Split used to the daily activity that a sudden change might incur a
80-20 12-6-3-1 100 78.95
12-6-3-1
chance of a heart attack. People engaged in exercising had
80-20 200 82.46
a lesser chance of being prone to a heart attack as
80-20 12-6-3-1 300 78.95 compared to the ones who didn't engage in exercising.
80 – 20 12-6-3-1 400 80.71 Angina was found to be a typical factor influencing heart
12-6-3-1
attack and it is advised for the people suffering from it to
80 -20 500 85.96
get a clinical check-up at the earliest.

C. Comparison and Prediction 56% of the patients displayed a chance of getting a

heart attack, and most of them tended to be men.
T ABLE X. MODEL BUILDING AND COMP ARISON
V II . DISCUSSION

Algorithm Accuracy
The development of accurate heart attack prediction
Logistic regression 0.912 models has significant implications for patient care. By
KNN 0.679 identifying high-risk individuals, doctors can implement
SVM 0.841 preventative measures, such as lifestyle changes or
Decision tree 0.824 medication, to reduce the likelihood of a heart attack
Random forest 0.825 occurring. This can lead to better health outcomes for
Bagging 0.842 patients and lower healthcare costs.
AdaBoost 0.824
GradientBoost 0.807 Furthermore, the development of accurate heart attack
XGBoost 0.812 prediction models has the potential to revolutionize the
Neural Network 0.859 healthcare industry. With more accurate and personalized
risk assessments, doctors can develop targeted treatment
All the algorithms were checked with multiple plans and improve patient outcomes. This can lead to a
parameters, the train-test split ratios, but a comparison 80:20 shift in the focus of healthcare from reactive to
train-test ratio was considered for all the models as it yielded preventative, which has significant implications for patient
the best accuracy in most cases. As seen from Table X., the care and healthcare costs.
logistic regression model gave the highest accuracy on
comparison and hence was further utilized to predict whether V II I. CONCLUSION
a person has a chance of having a heart attack or not. The Furthermore, heart attack prediction research is an
following array was given as input: essential area of study that has significant implications for
patient care and the healthcare industry. By developing
(41,0,1,130,204,0,0,172,0,1.4,2,0,2) accurate prediction models, researchers can identify high-
which resulted in the output that the person has a chance of risk individuals and implement preventative measures to
being prone to a heart attack. reduce the likelihood of heart attacks occurring. This has
the potential to improve patient outcomes and lower
healthcare costs while shifting healthcare focus from
V I. RESULT S
reactive to preventative notes.
The data were analyzed using different machine
learning algorithms ranging from logistic regression to I X. FUT URE SCOPE
Boosting, and the models were evaluated based on their The future scope for heart attack prediction research is
accuracy values which were noted. Table X. shows the broad and promising, potentially significantly improving
accuracy values for the respective algorithms. Apart from
public health by providing earlier identification and
this, the best model was also evaluated on another metric,
i.e., Recall, and the value was noted from the confusion intervention for those at risk for heart attacks. Here are
matrix (Fig 3.). some possible directions for future research:
Personalizing risk prediction: Risk factors for heart
The observed recall value was 93.33% attack can vary widely among individuals, and personalized
risk prediction can improve accuracy. Future research can
focus on identifying individual risk factors and developing
models that can accurately predict each person's heart attack
risk.

Developing mobile health technology: Mobile health

technology, such as wearable devices and mobile apps, can
collect real-time heart health data and provide personalized
risk prediction. Future research can focus on developing and
validating mobile health technology for heart attack
prediction.

Fig. 3. Confusion Matrix

979-8-3503-9720-8/23/$31.00 ©2023 IEEE

Authorized licensed use limited to: ST. JOSEPH ENGINEERING COLLEGE MANGALORE. Downloaded on March 11,2025 at 12:10:53 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Innovative Data Communication Technologies and Application (ICIDCA-2023)
IEEE Xplore Part Number: CFP23CR5-ART; ISBN: 979-8-3503-9720-8

[6] Mahmood, S. S., Levy, D., Vasan, R. S., & Wang, T . J. (2018).
Studying the impact of lifestyle interventions such as T he Framingham Heart Study and the epidemiology of
cardiovascular disease: a historical perspective. T he Lancet,
diet and exercise can have a significant impact on the risk 391(10173), 2317-2327.
of a heart attack. Future research can focus on studying the
impact of lifestyle interventions on heart attack risk and [7] Mueen, A., & Mohiuddin, A. K. (2020). Prediction of heart
developing personalized lifestyle interventions for disease using machine learning algorithms: a review. Journal of
Medical Systems, 44(6), 111.
individuals at risk.
[8] Rahimi, K., Bennett, D., Conrad, N., Williams, T . M., Basu, J.,
Dwight, J., ... & Prospective Studies Collaboration. (2014). Risk
X . LIMIT ATIONS OF ST UDY prediction in patients with heart failure: a systematic review and
analysis. J.A.C.C.: Heart Failure, 2(5), 440-446.
The study is limited in its scope as it has a limited
number of rows and columns. The presence of dynamic [9] Zhang, Y., Chen, L., Dai, J., Xu, Z., & Xu, X. (2020). Machine
learning for heart disease diagnosis: a systematic review. Frontiers
data instead of static data could prove to be more useful and in Cardiovascular Medicine, 7, 35.
relevant.
[10] Rathore, S., Gupta, V., Kumar, M., & Kumar, M. (2020). Heart
X I. AUTHOR’S CONT RIBUT ION attack prediction using machine learning algorithms: a systematic
review. Journal of Healthcare Engineering, 2020, 1-15.
All the authors contributed equally throughout the
paper. [11] Anwar, A., & Majid, M. (2021). An analysis of machine learning
models for heart attack prediction. International Journal of
Engineering and T echnology, 13(2), 119-126.

X II . REFERENCES [12] Hamdeh, M. A., Eldeib, A. M., & Alkandari, A. M. (2020). A

review of machine learning applications in cardiovascular disease
prediction. Journal of Healthcare Engineering, 2020, 1-11.
[1] Dey, D., Kumar, S., Mansuri, A., & Bajaj, V. (2020). Machine
learning-based prediction of heart attack using physiological and [13] Yildirim, Ö., T alo, M., Baloglu, U. B., Verkhodanov, D., &
demographic factors. Journal of Medical Systems, 44(7), 144. Rajendra Acharya, U. (2021). A review of machine learning-based
risk prediction models for heart failure. Expert Systems with
[2] T ison, G. H., Zhang, J., Delling, F. N., Deo, R. C., Dalal, D., Applications,166, 113572.
Aghayev, A., ... & Lee, K. (2019). Automated machine learning-
based electrocardiogram analysis for prediction of cardiovascular [14] Zhu, C., Li, X., Wang, Y., Zhang, H., & Zhou, J. (2021). An
events: a retrospective cohort study. T he Lancet Digital Health, overview of machine learning applications in cardiovascular
1(7), e352-e361. disease prediction. Biomedical Engineering Online, 20(1), 1-18.
[3] Xiong, Z., Gao, X., Zhou, B., & Liu, J. (2020). A neural network [15] Ahmed, A. M. (2021). Machine learning for heart disease
model for predicting heart attack based on demographic and clinical prediction: a comparative study. Journal of Ambient Intelligence
factors. B.M.C. Medical Informatics and Decision Making, 20(1), and Humanized Computing, 12(2), 1717-1731.
1- 10.
[16] Khemphila, A., & Sukchotrat, T . (2020). Predictive model for
[4] Alizadehsani, R., Habibi, J., Hosseini, M. J., & Khosravi, A. heart attack using machine learning algorithms. Journal of
(2020). Heart disease prediction using machine learning algorithms: Physics: Conference Series, 1548(1), 012045.
a review study. Journal of Biomedical Physics and Engineering,
10(1), 99-114. [17] Mohd, R., Ahmad, F., Yasin, N. M., & Faez, F. (2021). Heart
disease prediction using machine learning algorithms: a
[5] Desai, C. S., Sharif, B., Vaidya, O. S., & Shah, M. A. (2020). Heart comparative study. Journal of Ambient Intelligence and
disease prediction using machine learning algorithms: a systematic Humanized Computing, 12(8), 7987-8000.
review. Journal of Medical Systems, 44(7), 125.
[18] A. Kumar, A. Jain, B. Agarwal, M. Jain, P. Harjule, R.A. Verma.
(2021). Explainable Artificial Intelligence for Smart Cities, 1st
Edition, CRC Press, ISBN: 9781003172772,

979-8-3503-9720-8/23/$31.00 ©2023 IEEE

Authorized licensed use limited to: ST. JOSEPH ENGINEERING COLLEGE MANGALORE. Downloaded on March 11,2025 at 12:10:53 UTC from IEEE Xplore. Restrictions apply.