Heart Disease Prediction Using ML and Ensemble Models: A Comprehensive Analysis

The leading cause of death worldwide is still heart disease, highlighting the urgent need for improved diagnostics. This research included a range of ML techniques. Employing SVM, RF, DT, KNN, and LGBM to predict heart illness. The assessment metrics consist of F1 score, recall, accuracy, ROC, and precision. To ensure accurate analysis, data pretreatment techniques including reduction, integration, transformation, and purification were applied. Hyperparameter adjustment improved the model's performance even further. Among the models analyzed, Naïve Bayes performed the worst, with an F1 score of 98%, accuracy of 69.63%, precision of 93%, and recall of 98%. Three distinct datasets were employed in the study, and their strengths were combined to enhance model discrimination, reduce overfitting, and raise overall accuracy. We combined DT, LGNM, and LGBM to create a number of models. From RF we are able to achieve an accuracy of up to 98.89%, Precision 99.00%, Recall 98.00% and F1-Score 99.00%. These results show how well ensemble methods and data integration may enhance machine learning-based heart disease prediction abilities.