Trends in Blockchain-Assisted Federated Learning for Secure Medical IoT: A Review of Algorithms, Clinical Scenarios, and Deployment Barriers

Abstract

Background: Connected medical devices generate clinically valuable streams of physiological data, but centralised storage and processing raise persistent privacy, integrity, and accountability concerns. Federated learning (FL) and blockchain (BC) have emerged as complementary technical responses, the former keeping raw patient data on local hosts and the latter providing a tamper-evident audit substrate. Objective: This review consolidates recent advances in blockchain-assisted federated learning (BC-FL) for medical Internet of Things (IoT), focusing on algorithmic design choices, representative clinical use cases, and the operational barriers that slow real-world deployment. Methods: We synthesise findings from 50 peer-reviewed sources published primarily between 2017 and 2025, classify the corpus by methodological focus, and analyse reported performance trends, threat models, and clinical evaluation contexts. Results: BC-FL frameworks show measurable benefits in attack resistance, auditability, and cross-institutional collaboration; however, throughput, energy use, and regulatory ambiguity remain dominant constraints. We extract design patterns linking consensus selection, aggregation strategy, and privacy mechanism to specific clinical scenarios, and we summarise recurring deployment barriers. Conclusion: BC-FL is an increasingly mature paradigm for trustworthy medical AI, but pragmatic adoption in clinics will require lightweight consensus, standardised interoperability profiles, and transparent governance models, areas that we identify as priorities for the next research cycle.