FederatedRSF : Federated Random Survival Forests for Partially Overlapping Medical Data
Researchers have developed FederatedRSF, a Python package designed for federated random survival forests. This tool addresses the challenge of training predictive models on medical data from multiple institutions while adhering to privacy regulations and handling feature heterogeneity. FederatedRSF aggregates locally trained survival trees, allowing for inference without sharing raw patient data, and has demonstrated performance comparable to centralized training methods. AI
IMPACT Enables more robust and generalizable medical predictions by facilitating collaborative model training across institutions without compromising patient privacy.