I-Corps: Explanation-Based Auditing: Improving the Security of Electronic Medical Records

研究人员开发了一种基于解释的审计系统，通过从审计日志中过滤适当的访问来提高电子病历审计的效率，以便合规官员可以将精力集中在可疑行为上。该小组的审计日志分析方法消除了基于挖掘的逻辑解释的非可疑访问，并利用常见类型的缺失数据增强了这些解释。关键的见解是，电子病历(EMR)数据库已经存储了描述患者接受治疗的过程的信息(例如，预约、用药命令、诊断等)，这些信息可以用来解释为什么会发生访问。此外，审计系统自动发现这些解释，并将它们提交给合规官员批准，而不是要求合规官员手动说明访问应该发生的所有原因。以往的研究已经证实，该系统减轻了审计负担，以解释为基础的审计制度有可能减轻医院的审计和合规负担。首先，大型医院的合规办公室目前雇佣了数十人，每周手动处理和审查数十起投诉。有了拟议的技术，合规官员将只需要审查一小部分访问，大大减少了手动工作。先前的研究表明，由于临床或手术原因，超过94%的访问可以进行过滤。其次，医院患者可以要求对其医疗记录的所有访问进行记账，这被称为访问报告。目前，这些访问报告对患者来说是无法解释的，因为它们只列出访问医疗记录的员工，而不是他们是谁以及访问发生的原因。基于解释的审计系统通过附加描述访问的临床或操作原因的可解释文本描述来增强这些访问报告。第三，为了验证以解释为基础的审计制度的效率和效力，已经设立了一项试点试验，允许合规官员测试审计工具，并使用已知的误用实例来衡量其准确性。

Researchers have developed an explanation-based auditing system to improve electronic medical record auditing efficiency by filtering appropriate accesses from the audit log so compliance officers can focus their efforts on suspicious behavior. The team's approach to audit log analysis eliminates non-suspicious accesses based on mined logical explanations and enhances these explanations with common types of missing data. The key insight is that electronic medical record (EMR) databases already store information describing the process by which a patient is treated (e.g., appointments, medication orders, diagnoses, etc.), which can be used to explain why accesses occur. Moreover, instead of requiring compliance officers to manually specify all reasons for why accesses should occur, the auditing system automatically discovers these explanations and presents them to compliance officers for approval. Previous research has confirmed that the system reduces the auditing burden.The explanation-based auditing system has the potential to reduce the auditing and compliance burden for hospitals. First, compliance offices at large hospital currently employ tens of individuals to manually process and review tens of complaints a week. With the proposed technology, compliance officers will only need to review a small subset of the accesses, significantly reducing the manual effort. Previous research has shown that over 94% of all accesses can be filtered for clinical or operational reasons. Second, hospital patients can request an accounting of all accesses to their medical record, which are known as access reports. Currently, these access reports are un-interpretable for patients because they simply list the employees who accessed the medical record instead of who they are and why the access occurred. The explanation-based auditing system enhances these access reports by attaching interpretable text descriptions describing the clinical or operational reason for access. Third, to verify the efficiency and effectiveness of the explanation-based auditing system, a pilot trial has been set up at to allow compliance officers to test the auditing tool and measure its accuracy using known instances of misuse.