Development and Validation of an Open-Source Federated Learning Platform

Authors: Asti G, Apellaniz PA, Carota L, Merleau NSC, et al.

Published in: medRxiv (2025)

Citations: 2 citations

Abstract:
This paper presents the full development, implementation, and validation of an open-source federated learning platform for rare disease research. The platform implements differential privacy, secure aggregation, and model versioning, validated on multi-site hematological datasets. Experimental results confirm statistical performance equivalent to centralized approaches while providing formal privacy guarantees.