Enhancing Secure Patient Monitoring with Asynchronous Federated Learning: An Extension to WideResNet and FWBO-Based Fog-Blockchain Architecture

Abstract

Cardiovascular diseases continue to be among the leading causes of death globally. thus, effective and timely diagnosis is highly   necessary. This work proposes an intelligent detection framework, WideResNet FWBO, which will include classification with high accuracy using Wide Residual Networks together with the Fractional Wolf Bird Optimization technique. Empirical Mode Decomposition (EMD) will be utilized for pre-processing ECG signals, which will be further processed for feature extraction through MFCC, 1D-Local Ternary Patterns, and time-domain features. This shall be executed within a fog-assisted, blockchainsecured federated learning environment that will guarantee data privacy with low latency. The experiments conducted on ZENODOdatasets recorded 94.2% for accuracy.The proposed model reduces computational overhead and latency, improving responsiveness in remote monitoring by 30%, while ensuring security and scalabilityin a healthcare analytics environment.