Machine Learning
Coronary Artery Disease Prediction
Diagnoses coronary artery disease from routine clinical measurements. The popular Kaggle copy of the UCI Cleveland data has an inverted target, remapped codes and hidden missing values, so I audited and fixed it before modelling with nested cross-validation and an interpretable logistic regression.
Catches 27 of 28 disease cases on held-out patients (ROC AUC 0.936), with a Streamlit app that explains every prediction.

