Federated Learning Across Hospitals

Act as a ML systems architect. Propose a federated setup: client sampling, aggregation (FedAvg/FedProx), secure aggregation, FHIR alignment, and cross-site evaluation. Provide success metrics and rollback rules.