Federated Averaging Algorithm for Federated AI with Blockchain for Secure and Patient-Centric Healthcare

The present generation and future increasing adoption of AI in healthcare is constrained by critical challenges related to accuracy, privacy and preservation, and latency. Centralized-AI learning approaches require sensitive patient data to be shared, raising concerns about regulatory compliance and trust. To address these limitations, this paper presents a Federated Averaging (FedAvg) algorithm, and integrated FedAvg with Blockchain technology for secure and patient-centric healthcare. In this conference paper proposed healthcare institutions collaboratively train AI models using federated learning, where only local model updates are shared, while a permissioned blockchain ensures secure coordination, immutable logging, and trusted validation of model updates. Experimental evaluation demonstrates that the FedAvg with Blockchain achieved stable accuracy convergence across communication rounds, maintains high privacy preservation, and supports scalable collaboration with manageable latency as the number of participating healthcare clients increases. Finally, this work proved that uniquely combines FedAvg-based federated learning (FL) with Blockchain-based trust to achieve accuracy, privacy-preserving, auditable, and scalable patient-centric healthcare intelligence.