Machine Learning-Based Heart Disease Prediction: A Comparative Evaluation of Classification Methods

Cardiovascular diseases represent a major cause of mortality worldwide, highlighting the needfor reliable tools to support early diagnosis. This study evaluates the performance of supervisedmachine learning models for heart disease prediction using routinely collected clinical data from a publicly available dataset. Six classification algorithms are investigated, ranging from linear models to ensemble-based and neural approaches.A reproducible experimental framework is adopted, including data preprocessing, feature encoding, feature engineering, and normalization, followed by a stratified 80/20 train–test split. Model evaluation is conducted using clinically relevant metrics, with particular emphasis on recall in order to limit false negative predictions in a screening context.The results indicate that all models achieve comparable performance, with accuracy values between 86.96% and 88.04%. Linear models such as logistic regression and support vector machines exhibit high sensitivity, while ensemble-based methods, particularly random forest, provide the most balanced performance, reaching an accuracy of 88.04% and a recall of 92.16%.Overall, these findings demonstrate the potential of machine learning techniques to support heart disease prediction and underline the importance of evaluation strategies aligned with clinical objectives. Although the study relies on a single dataset, the proposed framework offers a basis for further validation on larger and more diverse clinical cohorts.