Skip to content
InfoResearchPeer-reviewed

Secure and Resilient Clustered Federated Learning for Web-Enabled Healthcare Analytics Using Lightweight Blockchain and Adaptive Model Selection

Published
Record updated
View JSON

Summary

Researchers propose Blockchain-based Clustered Federated Learning (BCFL), a cluster-based federated learning framework for web-enabled healthcare analytics. It replaces a trusted central aggregator with a lightweight permissioned ledger that stores compact provenance metadata on-chain while keeping model parameters off-chain. Experiments on real-world healthcare data and other benchmarks report improved model performance and lower attack success rates compared with several FL baselines.