A modern, explainable breast cancer prediction demo
We converted a notebook into a reproducible training pipeline with model comparison, stacking, and optional deep-learning. Deployable with FastAPI and demoed with this Next.js frontend. Visuals are generated on-demand from the API, no static artifacts required.
Production-ready pipeline
Imputer → Scaler → SelectKBest → candidate models + GridSearchCV + stacking
On-demand plots
ROC, PR, confusion, feature importances and SHAP are streamed as images
Explainable
SHAP-based explanations (when available) or local contributions inlined
Awareness Spotlight
Why Early Detection Matters — Short Overview
Embedded via YouTube privacy-enhanced mode to reduce tracking.