<![CDATA[Heart disease remains a major contributor to global mortality, highlighting the importance of effective early detection systems that can assist both clinicians and general users. This study develops a heart disease prediction model based on the XGBoost algorithm and deploys it within a web-based application to enhance accessibility and practical usability. The research uses a dataset of 918 instances containing 12 demographic and clinical features commonly associated with cardiovascular risk. Pearson correlation analysis was performed to assess feature relevance, revealing that ExerciseAngina, Oldpeak, ST_Slope, Age, and MaxHR exhibit the strongest correlations with the HeartDisease outcome. These findings align with established clinical evidence on exercise-induced angina, ST-segment depression, and cardiac functional capacity. Following preprocessing and feature encoding, the XGBoost model was trained and evaluated. The model achieved strong predictive performance, with 88.26% accuracy, 88.32% precision, 91.66% recall, an F1-score of 89.96%, and an ROC-AUC of 0.93. The results demonstrate that XGBoost effectively discriminates between positive and negative cases and provides a good balance between sensitivity and precision. To enable real-world applicability, the final model was deployed on a Flask backend and integrated into a web application that allows users to input clinical parameters and receive real-time predictions. System testing confirmed that the application accurately delivers outputs and functions reliably across different input conditions. Overall, this study shows the feasibility of combining machine learning with web technologies to support early, accessible heart disease screening. Future work will involve usability testing and validation using real patient data to further strengthen the system’s clinical relevance.]]>
Copyrights © 2026
