Scholarlink

<p>E-health systems increasingly interconnect IoT devices, edge resources, and cloud backends, making secure and accountable access to electronic health records (EHRs) both critical and challenging. Static, attribute-only policies struggle to capture temporal and relational dependencies in access behaviors; moreover, conventional auditing is coarse-grained and brittle when the workload or data changes. We propose a context-aware access control framework that treats audit logs as a temporal, heterogeneous graph and scores each request with a Temporal Heterogeneous Graph Transformer (THGT). By capturing how users, devices, and resources interact over time, the model produces calibrated, context-specific risk estimates that outperform static, attribute-only rules. Decisions are governed by an adaptive soft threshold that triggers step-up verification only for borderline cases, balancing security and usability without imposing unnecessary friction on routine access. To preserve performance and verifiability, payloads remain off-chain. At the same time, model/threshold versions, as well as decision digests, are anchored on a permissioned blockchain, providing immutable, end-to-end integrity and accountability. Together, these elements yield reliable admissions, lower latency, and a transparent audit trail, making them suitable for EHR environments with dynamic workloads. In simulations reflecting IoT-enabled healthcare, and using the reference EHR dataset, our method reduces the false rate by 8%, increases the access rate by 7%, lowers the access time by 130&#xa0; ms, decreases the processing time by 120&#xa0; ms, and improves the throughput by 7% versus the strongest baselines, these gains persist across various loads and user populations, indicating a favorable trade-off between accuracy and efficiency, as well as robust scaling.</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

A graph transformer and blockchain synergy approach for healthcare privacy

  • 期刊论文

  • 发表日期：

    2026年7月6日

查看全文
  • Abbas Hassan Pour Askari,
  • Amid Khatibi Bardsiri,
  • Mokhtar Mohammadi Ghanatghestani

摘要

E-health systems increasingly interconnect IoT devices, edge resources, and cloud backends, making secure and accountable access to electronic health records (EHRs) both critical and challenging. Static, attribute-only policies struggle to capture temporal and relational dependencies in access behaviors; moreover, conventional auditing is coarse-grained and brittle when the workload or data changes. We propose a context-aware access control framework that treats audit logs as a temporal, heterogeneous graph and scores each request with a Temporal Heterogeneous Graph Transformer (THGT). By capturing how users, devices, and resources interact over time, the model produces calibrated, context-specific risk estimates that outperform static, attribute-only rules. Decisions are governed by an adaptive soft threshold that triggers step-up verification only for borderline cases, balancing security and usability without imposing unnecessary friction on routine access. To preserve performance and verifiability, payloads remain off-chain. At the same time, model/threshold versions, as well as decision digests, are anchored on a permissioned blockchain, providing immutable, end-to-end integrity and accountability. Together, these elements yield reliable admissions, lower latency, and a transparent audit trail, making them suitable for EHR environments with dynamic workloads. In simulations reflecting IoT-enabled healthcare, and using the reference EHR dataset, our method reduces the false rate by 8%, increases the access rate by 7%, lowers the access time by 130  ms, decreases the processing time by 120  ms, and improves the throughput by 7% versus the strongest baselines, these gains persist across various loads and user populations, indicating a favorable trade-off between accuracy and efficiency, as well as robust scaling.