<![CDATA[Heart disease is one of the leading causes of death worldwide, requiring an accurate risk prediction system to support early detection and clinical decision-making. This study aims to develop a Personalized Cardiovascular Composite Score (PCCS) as a composite cardiovascular risk score and to compare the performance of the Logistic Regression and Random Forest models in predicting heart disease. The dataset used consists of numerical and categorical clinical variables representing cardiovascular risk factors. The research stages included data exploration, preprocessing using one-hot encoding and standardization techniques, the creation of a Risk Multiplier based on clinical factors, and the calculation of the PCCS as a composite risk index. The dataset was then split using a stratified split method with an 80:20 ratio for model training and testing. Performance evaluation was conducted using the metrics of accuracy, precision, recall, F1-score, confusion matrix, and Area Under the Curve (AUC) on the ROC curve. Random Forest achieved a nominally higher AUC than Logistic Regression, but that this advantage was not statistically significant, indicating comparable discriminative performance between the two algorithms. However, Logistic Regression still offers an advantage in terms of clinical interpretability. These findings come from a single public retrospective dataset. They show that a clinically weighted composite feature can be built and used with both classifiers, but they do not by themselves establish clinical readiness; external validation and calibration assessment are required before any use in risk-based clinical decision support.]]>
