Background: Cardiovascular disease (CVD) remains a leading cause of morbidity and mortality worldwide. Traditional risk assessment methods often lack the predictive power needed for early and effective intervention. This study explores the potential of a machine learning-based framework to enhance the accuracy of CVD risk prediction.

Methods: We developed a specialized framework utilizing supervised learning algorithms to predict heart disease severity. The study leveraged the publicly available UCI Heart Disease Dataset, which contains 14 clinical and demographic attributes. We preprocessed the data and applied feature selection techniques before training and evaluating four models: Logistic Regression, Decision Trees, Random Forests, and XGBoost. The performance of each model was rigorously evaluated using standard metrics, including accuracy, precision, recall, and F1 score.

Results: A comparative analysis revealed that XGBoost consistently demonstrated superior performance among the tested algorithms. The XGBoost model achieved the highest accuracy, at 62.5%, indicating its strong capability in identifying at-risk patients. The other models showed varied performance, underscoring the importance of model selection for this task.

Discussion: The findings confirm that machine learning, and specifically the XGBoost algorithm, can effectively analyze complex clinical data to predict cardiovascular disease risk. This framework holds promise as a powerful clinical decision-support tool. Future work should focus on validating the framework with larger datasets and exploring its integration into clinical practice.