A blood and marrow transplant is a procedure that replaces unhealthy blood-forming cells with healthy ones. It typically involves using blood-forming cells donated by someone else instead of one's own blood-forming cells. The goal of this project is to predict transplant survival rates for post Bone Marrow Transplant patients.

We implemented and finetuned four models including Cox Proportional Hazard Model, XGBoost AFT, Survival Random Forest, and CatBoost AFT. To improve model performance, we hybridized each of the four models with an extra logistic or random forest stratification.

Our dataset comes from a Kaggle competition: https://www.kaggle.com/competitions/equity-post-HCT-survival-predictions/.