This paper presents a web-based cardiovascular disease (CVD) risk prediction system that combines supervised machine learning with an accessible user interface. An XGBoost classifier is trained on a cleaned Kaggle CVD dataset using eleven low-cost clinical and lifestyle features. The model achieves competitive performance (accuracy ~0.73, ROC-AUC ~ 0.80) and is deployed via a Streamlit application that provides probability-based risk categories for preliminary self-screening, illustrating an end-to- end pipeline from data preprocessing to cloud deployment.
Singh et al. (Tue,) studied this question.