CVD Risk Predictor — AI-Powered Patient Treatment Recommendation System

A full-stack machine learning web application that predicts cardiovascular disease risk from 18 patient health indicators and generates tiered treatment recommendations (T1–T5). Built on a Gradient Boosting model trained on 308,854 records with SMOTE for class imbalance, isotonic calibration (Brier: 0.065), and an AUC of 0.827. Features SHAP explainability, fairness auditing, and EU AI Act compliance. Extended from an MSc dissertation with significant technical improvements.