Scholarlink

<p>Blockchain can secure and verify electronic health records (EHRs) for multi-institution healthcare systems, but Layer-1 storage costs and throughput limitations make full on-chain EHR storage impractical. A new model is proposed named FZRP (Federated-ZK-Rollup Pipeline). It is a hybrid methodology combining Federated Learning (FL), off-chain storage (IPFS), zk-rollup batching with adaptive batch sizing, and parallel proof pipelines to minimize per-record transaction cost while preserving auditability and privacy. Using a synthetic dataset of 50,000 EHRs, it quantifies cost reductions under realistic assumptions and demonstrate orders-of-magnitude per-record savings. A formal cost model, latency and security analyses, and sensitivity studies are provided. The experimental evaluation demonstrates that adaptive batching significantly reduces per-record transaction cost under conservative Layer-1 cost assumptions to as low as $0.000024, achieving over 99.999% cost reduction while maintaining scalability and privacy. The limitations, regulatory considerations, and paths for future work are discussed.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Efficient design of blockchain-based electronic health record systems with federated learning, off-chain storage, and zk-rollup parallelism for minimizing transaction costs

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月1日

查看全文
  • Abhinav Raghav,
  • Aanjey Mani Tripathi,
  • Rajesh Kumar Chaudhary

摘要

Blockchain can secure and verify electronic health records (EHRs) for multi-institution healthcare systems, but Layer-1 storage costs and throughput limitations make full on-chain EHR storage impractical. A new model is proposed named FZRP (Federated-ZK-Rollup Pipeline). It is a hybrid methodology combining Federated Learning (FL), off-chain storage (IPFS), zk-rollup batching with adaptive batch sizing, and parallel proof pipelines to minimize per-record transaction cost while preserving auditability and privacy. Using a synthetic dataset of 50,000 EHRs, it quantifies cost reductions under realistic assumptions and demonstrate orders-of-magnitude per-record savings. A formal cost model, latency and security analyses, and sensitivity studies are provided. The experimental evaluation demonstrates that adaptive batching significantly reduces per-record transaction cost under conservative Layer-1 cost assumptions to as low as $0.000024, achieving over 99.999% cost reduction while maintaining scalability and privacy. The limitations, regulatory considerations, and paths for future work are discussed.