FederatedRSF : Federated Random Survival Forests for Partially Overlapping Medical Data

In three linesFederatedRSF is a Python package implementing federated random survival forests for multi-center survival prediction without sharing raw patient data. The system handles feature-space heterogeneity (different covariates across sites) by redistributing only compatible trees. Evaluation on GBSG2 breast cancer cohort shows performance comparable to centralized training.

Summary generated by Claude — human-verified