Enhancing Health Data Sharing through Blockchain and IPFS Integration with Machine Learning Based Optimization

This preprint presents a decentralized framework that integrates blockchain technology with the InterPlanetary File System (IPFS) to enable secure, interoperable sharing of Electronic Health Records (EHRs). We address the limitations of centralized health data systems—privacy risks, security weaknesses, and poor cross-platform compatibility—by proposing a dual-layer architecture: (1) an abstract service layer that unifies data access across heterogeneous blockchain platforms (e.g., Ethereum and Hyperledger Fabric), and (2) a patient-controlled access mechanism leveraging smart contracts and machine-learning–based anomaly detection. Using synthetic datasets, experimental evaluation shows a 15% reduction in data exchange latency and an improvement in transaction success rate (e.g., from 95% to 98%) compared to non-optimized routing. The integration of Random Forest for routing optimization and Isolation Forest for anomaly detection enhances both performance and security, achieving accuracies of 95% and 92%, respectively. Overall, the study demonstrates the feasibility of scalable, secure, and patient-centric health data exchange in a decentralized environment.