HEART DISEASE PREDICTION USING MACHINE LEARNING CLASSIFIERS WITH VARIOUS BALANCING TECHNIQUES

Aug 5, 2026

Upload by Ezra

Heart disease or cardiovascular illness is a major cause of global mortality, and predicting it from clinical data remains challenging due to class imbalance and noisy information. This study compares oversampling strategies—SMOTE, SMOTE-ENN, and ADASYN—integrated with six machine learning classifiers: Logistic Regression, SVM, KNN, Decision Tree, Random Forest, and Gradient Boosting. Model performance is evaluated using the accuracy metric. Results show ADASYN delivers the best performance and increases predictive accuracy for heart failure prediction.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Introduction

Primary contribution

Literature Survey

Vol. 06, No. 4 (2024) 1871-1878, doi: 10.24874/PES.SI.25.03A.017 Proceedings on Engineering Sciences [www.pesjournal.net](www.pesjournal.net) HEART DISEASE PREDICTION USING MACHINE LEARNING CLASSIFIERS WITH VARIOUS BALANCING TECHNIQUES Uzama Sadar Parul Agarwal 1 Received 05.01.2024. Suraiya Parveen Received in revised form 08.02.2024. Geetika Dhand Accepted 11.03.2024. Kavita Sheoran UDC – 004.85 Keywords: Machine Learning, Heart Disease, Prediction Model, Balancing Techniques, Classification Algorithms A B S T R A C T Heart disease or Cardiovascular illness is the most prevalent cause of mortality globally. The challenge of predicting heart illness using clinical data analytics is considerable. Machine learning (ML) has been extensively used in the medical domain for disease prediction. This work performs a comparative analysis of various oversampling methods like the Synthetic Minority Oversampling technique (SMOTE), Synthetic Minority Oversampling Technique with Edited Nearest Neighbor (SMOTE-ENN), and Adaptive Synthetic Sampling Approach (ADASYN) Algorithm used with ML classifierson imbalanced heart failure prediction dataset. Six ML classifiers are analyzed in the study Logistic Regression (LR), Support Vector Machine (SVM), K Nearest Neighbor (KNN), Decision Tree (DT), Random Forest (RF), and Gradient Boosting (GB). The accuracy metric is used to measure the model’s performance The result depicts that the ADASYN technique performs better for the given dataset and increases the accuracy of the classifier in heart failure prediction. © 2024 Published by Faculty of Engineering 1. INTRODUCTION The heart is considered the central organ of a human being. Cardiovascular Disease (CVD) has been described as the most severe and fatal disease worldwide. A significant risk and burden are being placed on the world’s healthcare system by the rise in cardiovascular illness with a high death rate. A World Health Organization (WHO) research states that 17.9 million individuals worldwide die from CVD every year (World Health Organisation, 2020) . If appropriate action is not taken, that number will rise to 22 million in 2030 (Nagavelli, Samanta & Chakraborty, 2022) . CVD is a disease of the blood vessels and Heart. Blockage of arteries that supply oxygen and blood causes heart disease. According to the report NHIS (National Health Interview Survey), the five most prevailing symptoms of Heart attack are (Fang, Luncheon, Ayala, Odom, & Loustalot, 2019) Chest pain, Excessive Sweating, Dizziness, Fatigue, and Pain in the arm and jaw. Angiography is still the most reliable technique for identifying cardiac anomalies, but it is costly and requires a high degree of technical competence (Phegade et al., 2019) Therefore, clinicians are supported by a Machine learning-based predictive decision support system to reduce the death rate and improve the decision-making process (Rani, Kumar, Ahmed & Jain, 2021). Nowadays, Artificial Intelligence subparts Machine Learning and Deep Learning have been extensively used in clinical decision-making and assist in disease detection and prediction (Ramana et al., 2022). With the development of computers, the healthcare industry now has access to boundless amounts of clinical data in MRI, sensor, and electronic health record data (Dhand, Sheoran, Agarwal & Biswas, 2022) It is challenging for medical professionals to extract pertinent information from this data. 1.1 The paper's primary contribution is explained as follows 􀁸 This work performs a comparative analysis between oversampling methods. 􀁸 Three oversampling methods namely SMOTE, SMOTE-ENN, and ADASYN are integrated with 6 states of the art algorithm LR, SVM, KNN, DT, RF, and GB. 􀁸 The experimentation was done on the imbalanced Heart Failure Dataset. 􀁸 Accuracy metric was used for performance evaluation and results show that ADASYN performs better for the given dataset. The remaining part of the paper is struc

FAQs

Which oversampling techniques are compared in the study?

The study compares SMOTE, SMOTE-ENN, and ADASYN to handle imbalance in the dataset.

Which machine learning classifiers are evaluated?

Six classifiers are analyzed: Logistic Regression, Support Vector Machine, K Nearest Neighbor, Decision Tree, Random Forest, and Gradient Boosting.

What oversampling method gives the best results and how is performance measured?

ADASYN performs best on the given imbalanced heart failure dataset, and performance is measured using the accuracy metric.

Aug 5, 2026

Upload by Ezra