Optimizing Proof-of-Work for Secure Health Data Blockchain Using Compute Unified Device Architecture

Abstract

We present a GPU-accelerated proof-of-work (PoW) blockchain design tailored for secure healthcare data management. Our CUDA-optimized PoW achieves significantly higher throughput and lower block-formation latency compared to CPU mining, making blockchain practical for high-volume health records. We benchmark against standard platforms: Bitcoin (SHA-256 PoW), Ethereum (legacy PoW), and Hyperledger Fabric (permissioned consensus) to quantify performance gains. Empirical tests show GPU-accelerated encryption (AES-CTR) processes large health-record payloads in under 1 second, while our PoW mining throughput improves by ~20–50% relative to unaccelerated baselines. We also evaluate end-to-end encryption latency and discuss privacy trade-offs: lightweight AES yields minimal delay, whereas fully homomorphic methods are impractical in a permissionless setting. We explicitly address regulatory compliance: personal health data are stored off-chain (e.g. in IPFS), preserving the “right to erasure” via deletion of off-chain records, and we implement strict access controls to meet HIPAA security rules. The design includes validator selection rules that limit Sybil attacks by requiring costly work (or stake) and supports post-quantum cryptographic agility (e.g. Falcon signatures). We define our research question (“Can CUDA-accelerated PoW enable a high-performance yet compliant health data blockchain?”) and hypothesize that GPU parallelism will yield orders-of-magnitude speed-ups. Results confirm our hypothesis: throughput and latency are dramatically improved while preserving data privacy and compliance. This work makes a comprehensive contribution by detailing implementation methods, performance benchmarking, and analysis of security and legal requirements in a unified blockchain framework for healthcare.