Heart Disease Prediction using Optimized Machine Learning Models

Heart disease remains the leading cause of mortality worldwide, emphasizing the importance of early prediction, which is often challenging. This study proposes an optimized machine learning framework to evaluate the performance of various algorithms in predicting heart disease. Five algorithms-logistic regression (LR), decision tree (DT), random forest (RF), XGBoost (XGB), and multilayer perceptron (MLP)-were applied to the Cleveland dataset. Hyperparameters were optimized through Bayesian optimization and randomized grid search. Results indicated that RF and XGB achieved the highest accuracy ($90.0 \%$) and the largest AUC values (0.960 and 0.950, respectively). Logistic regression and decision trees demonstrated comparable accuracy ($89.0 \%$), while multilayer perceptron showed comparatively limited performance ($88.0 \%$). This work highlights the potential of machine learning techniques in the early detection of heart disease.