Comparative Study of Heart Disease Classification Based on Traditional and Ensemble Models

Heart disease poses a serious and significant risk to human health. Timely and accurate diagnosis is essential for effective prevention and treatment, which can greatly reduce mortality rates. Traditional diagnostic methods often face limitations in accuracy and scalability, highlighting the need for advanced computational approaches. Machine learning has emerged as a powerful tool to address these challenges by analyzing complex patterns in medical data. In this paper, we propose a novel model combining various machine learning techniques to enhance the detection of heart disease. Our approach focuses on creating a robust training dataset through comprehensive data gathering, preprocessing, and transformation, integrating data from multi-ple sources: Cleveland, Long Beach VA, Switzerland, Hungarian, and Statlog datasets. We developed hybrid classifiers, including AdaBoost, AbaBoost Bagging Method (ABBM), Gradient Boost, Random Forest, Random Forest Bagging Method (RFBM), Majority Voting, XGB classifier, and Stacked Generalization. These were built by integrating conventional classifiers with advanced bagging and boosting techniques to improve prediction performance. Evaluation metrics such as Accuracy, Sensitivity, Precision, and Fl-Score were employed to rigorously assess the models. Among these, the proposed Stacked Generalization model demonstrated superior performance, achieving the highest accuracy of 95.38 %. The findings underscore the model's potential in effectively identifying individuals at risk of heart disease, offering a significant contribution to early detection and public health initiatives.