The rapid adoption of Internet of Medical Things (IoMT) in Federated Learning (FL)-enabled Smart Healthcare 5.0 raises pressing concerns regarding privacy, security, and real-time threat detection. FL remains vulnerable to training-phase attacks, parameter breaches, and aggregation threats, with the central server posing risks of inference, poisoning, and single-point failure. To address the challenge of single point failure and lack of efficient decentralized trust mechanism, we propose a novel framework that integrates Split Federated Learning (SFL) with Blockchain. SFL uses hybrid Deep Learning (DL) model which is divided between the IoMT and Edge layers by the architecture: Edge-based BiLSTM networks identify threats and temporal patterns, while lightweight CNN at the IoT layer extract spatial features from patient data. While SFL facilitates safe decentralized training, Blockchain uses the Practical Byzantine Fault Tolerance (PBFT) consensus mechanism to guarantee tamper-proof integrity, trust, and authentication. This method improves decentralized trust, protects privacy, and lessens data leakage. The framework is a robust and real-time intrusion detection solution for next-generation smart healthcare, as demonstrated by experiments on ToN-IoT and IoT healthcare datasets, which show improved privacy, robustness against attacks, block commit rate (450 b/s), and reduced consensus time (250 ms) in addition to superior accuracy (99.95%).
Baihan et al. (Tue,) studied this question.