Safeguarding Patient Data-Sharing: Blockchain-Enabled Federated Learning in Medical Diagnostics

Somaraju Arun1

PG Scholar,

Dept of CSE Sphoorthy Engineering College Hyderabad, India arunsomaraju1@gmail.com

Dr. Kaja Mastan2

Assistant Professor,

Dept of CSE Sphoorthy Engineering College Hyderabad, India

Mrs. D. Srilatha Reddy3 Assistant Professor,

Dept of CSE Sphoorthy Engineering College Hyderabad, India

Abstract- The sharing of patient medical data for diagnostic analysis is difficult because healthcare records and medical images are highly sensitive and must remain protected against unauthorized access, leakage, and tampering. This research proposes a Blockchain-enabled Federated Learning (BCFL) system for secure medical diagnostics. In the proposed framework, hospitals and healthcare nodes train diagnostic models locally, while only encrypted model updates are shared through a blockchain-assisted environment. Blockchain provides immutable logging, transparent access control, smart contract-based governance, and trustworthy coordination among participants. Federated Learning protects patient privacy by avoiding centralized data collection and allowing collaborative training across distributed medical datasets. The system is designed for lung disease detection using the NIH Chest Xray dataset containing 112,120 X-ray images and supports multi-class diagnosis, including cases where multiple diseases may occur together. The proposed model achieved a diagnostic accuracy of 92.86%, latency of 43.518625 MS, and throughput of 10034017 bytes/s. Security evaluation against major cyber threats shows robustness close to 87%, demonstrating that BCFL can improve privacy-preserving, secure, and scalable healthcare diagnostics.

Keywords- Blockchain, Federated Learning, Medical Diagnostics, Patient Data Privacy, Healthcare Security, Lung Disease Detection, Smart Contracts, Cybersecurity, NIH Chest Xray Dataset