Privacy-Preserving Health Data Sharing Using Secure MPC

Abstract

Collaborative data analytics holds tremendous potential for healthcare research, but privacy regulations and ethical obligations limit data sharing between institutions. Traditional anonymization methods remain vulnerable to re-identification attacks, necessitating cryptographic techniques that enable computation without exposing raw data. We present a full-stack platform for privacy-preserving health data exchange that combines Secure Multi-Party Computation (SMPC), Homomorphic Encryption (HE), and Differential Privacy (DP). The system supports eighteen analytical tasks spanning descriptive statistics, regression, survival analysis, and federated machine learning. Comprehensive testing with simulated multi-institutional scenarios demonstrates that secure computations achieve accuracy within 0.02% of plaintext baselines while scaling linearly with dataset size. The platform integrates compliance features, real- time monitoring, and access control mechanisms aligned with HIPAA and GDPR requirements. Code and data are publicly available for reproducibility. This work bridges the gap between cryptographic theory and production healthcare systems, pro- viding a foundation for multi-institutional collaboration without compromising patient privacy.