Predicting Heart Diseases Using Machine Learning and Different Data Classification Techniques

Heart Diseases remains one of the foremost causes of death worldwide, making timely and accurate discovery a critical necessity.In this research, machine learning (ML) approaches were employed to improve predictive performance using a dataset compiled from both private and publicly accessible sources.Three distinct feature selection techniques-Chi-square, Analysis of Variance (ANOVA), and Mutual Information (MI)-were applied, producing three corresponding feature groups named SF-1, SF-2, and SF-3.To evaluate predictive capability, ten machine learning classifiers were tested: Naive Bayes, Support Vector Machine (SVM), XGBoost, AdaBoost, Bagging, Decision Tree, K-Nearest Neighbor (KNN), Random Forest, Logistic Regression, and a Voting Classifier that integrated multiple models.Classifier performance was measured through cross-validation, while Synthetic Minority Oversampling Technique (SMOTE) was implemented to correct class imbalance in the dataset.Among the models, the Voting Classifier delivered accuracies of 86%, 82%, and 82% across the different feature groups.More notably, a Stacking Classifier built using a combination of Boosted Decision Tree and ExtraTree with LightGBM achieved a perfect accuracy of 100%.These findings highlight the effectiveness of ensemble methods in significantly enhancing heart disease prediction.