IEEE J Biomed Health Inform. 2022 May;26(5):1917-1927. doi: 10.1109/JBHI.2021.3123643. Epub 2022 May 5.
The digitization of Electronic Medical Record (EMR) provides potential access to a wealth of medical information, but also presents new challenges in privacy-preserved EMR exchanging and sharing. In this paper, we propose a blockchain-based smart healthcare system with fine-grained privacy protection for reliable data exchanging and sharing among different users. We design a blockchain-enabled dynamic access control framework combined with Local Differential Privacy (LDP) strategies to provide the attribute-based privacy protection in transaction workflow. We design four types of smart contracts in the framework to meet the requirements of anonymous transaction, dynamic access control, beneficial matching decision, and evaluation of published data in an open network. To satisfy fine-grained privacy protection, we classify sensitive attributes of EMRs into different levels and set differential privacy budgets to randomize attributes before data publishing. Also, we design data quality function to depict the disturbance incurred by LDP-based privacy preferences at the requester view, and present appropriate many-to-many matching decisions among participants for beneficial transactions. Finally, we develop a prototype system and test our approach using 200,000 real-world EMRs. Experimental results show that the proposed privacy-preserved scheme can make stable and reliable transactions between EMR publishers and requesters. The prototype system achieves individual-centric privacy configuration at the patient site, while providing error-guaranteed statistics at the requester site. Additionally, the access control policies, logs of anonymous transaction are kept in the blockchain to provide system-level traceability.
电子病历(EMR)的数字化为获取丰富的医疗信息提供了可能,但也给隐私保护的 EMR 交换和共享带来了新的挑战。在本文中,我们提出了一种基于区块链的智能医疗系统,具有细粒度的隐私保护,可在不同用户之间可靠地交换和共享数据。我们设计了一个基于区块链的动态访问控制框架,结合局部差分隐私(LDP)策略,为交易流程中的基于属性的隐私保护提供支持。我们在框架中设计了四种智能合约,以满足匿名交易、动态访问控制、有益匹配决策和开放网络中发布数据评估的要求。为了满足细粒度的隐私保护,我们将 EMR 的敏感属性分为不同的级别,并为数据发布前的属性设置差分隐私预算,以进行随机化。此外,我们设计了数据质量函数来描述 LDP 隐私偏好在请求者视图中引起的干扰,并为有益交易提供参与者之间的适当的多对多匹配决策。最后,我们开发了一个原型系统,并使用 20 万份真实的 EMR 对我们的方法进行了测试。实验结果表明,所提出的隐私保护方案可以在 EMR 发布者和请求者之间实现稳定可靠的交易。原型系统在患者端实现了以个体为中心的隐私配置,同时在请求者端提供了有错误保证的统计数据。此外,访问控制策略和匿名交易日志都保存在区块链中,以提供系统级的可追溯性。