Heart Disease Risk Prediction

Co-developed an end-to-end ML pipeline on the UCI dataset (303

instances, 14 attributes), integrating PCA and Feature Selection to isolate

key health risk factors.

•

Led the hyperparameter tuning phase using GridSearch &

RandomizedSearch, improving classifier accuracy to ~87% (estimate) and

reducing false negatives for critical predictions.

•

Deployed the final model via Streamlit and ngrok, creating a live web

interface that enables real-time risk assessment from user inputs.