Multimodal Outcome Prediction from Linked Multi-Source Data
Multimodal XGBoost pipeline integrating structured clinical data with text-derived features across 175,000+ records.
A multimodal XGBoost pipeline integrating structured clinical data with text-derived features, predicting outcomes across 175,000+ records with a ROC-AUC of 0.86. Includes geospatial and temporal feature engineering built on Azure.
Tech stack: Python, XGBoost, SHAP, NLP feature extraction, geospatial/temporal feature engineering, Azure
GitHub: [GITHUB LINK TBD]