A Privacy-Preserving Healthcare System Using Secure Federated Learning and Blockchain for Industry 5.0
Anita Murmu, Naween Kumar, Akansha Singh, Krishna Kant Singh · IEEE Communications Standards Magazine · 2025
Federated learning (FL) is a scalable machine learning (ML) paradigm that works alongside the Internet of Medical Things (IoMT) through privacy-preserving mechanisms to handle medical imaging problems. FL provides a system through which multiple clients can train models to share aggregated parameters without exchanging raw information. IoMT enables medical device connections for health improvement, and FL uses 5G networks with edge computing to fulfill Industry 5.0 principles. Security and privacy issues persist as the existing methods are used to handle sensitive medical information. The paper presents FedAvg for medical images that combines the CNN-FedAvg protocol security with the CusDEFL deep learning framework data protection. The proposed work creates secure keys by implementing a 2D Chaotic Sine Map (2D-CSM) to protect medical images. The simulation tests revealed superior data protection, higher precision, and better data privacy than the baseline methods.