Safeguarding Clinical Data Privacy through Block chain Integrated For Multi Disease Detection

Abstract

Medical healthcare systems are increasingly leveraging artificial intelligence to manage large scale clinical data and improve disease diagnosis; however, traditional machine learning methods often compromise patient privacy due to centralized data sharing. This study proposes a secure and privacy preserving framework that integrates Blockchain with Federated Learning, enabling multiple hospitals to collaboratively train models without sharing raw patient data. Each hospital trains a local RESNET based model on its dataset and shares only encrypted model parameters, which are securely stored and aggregated via Blockchain to prevent tampering. The framework is validated on 15 lung diseases using the NIH Chest X ray dataset, achieving an high accuracy, latency of 43.51 ms, and throughput of 10,034,017 bytes/s, demonstrating efficiency and scalability. The system also resists major cyber threats, including Double Spending, Transaction Malleability, and Endpoint Compromise attacks, highlighting its robustness. Overall, this work provides a practical and secure implementation of Blockchain
enabled Federated Learning for medical diagnostics while safeguarding patient data privacy