Enhanced Heart Disease Prediction Through Optimized Ensemble Random Forest Model

Nowadays, the necessity for precise early detection of heart disease is critical with the conventional methods. However, challenges such as handling imbalanced datasets, ensuring model interpretability, and avoiding overfitting persist in predictive models. This paper assesses the efficacy of various machine learning algorithms by analyzing a vast dataset of cardiovascular health indicators. The approaches employed consist of classifiers such as Logistic Regression, K- Nearest Neighbors, Support Vector Machines, Naive Bayes, Decision Trees, Random Forests, Gradient Boosting, Stochastic Gradient Descent, and AdaBoost. Finally, these algorithms will be evaluated using performance indicators such as F1-score, ROC curves, DET curves, accuracy, precision, recall, and recall. The results indicated that the Decision Tree and Random Forest classifiers are the most adequate, each achieving an accuracy of 98.54% and near-perfect AUC scores of 0.99 and 1.00, respectively. This comparative analysis highlights the superiority of ensemble learning techniques in making robust predictions, thereby significantly contributing to the early detection and management of heart disease.