Transforming Patient Safety Event Data into Actionable Insights through Advanced Analytics

摘要 提出的研究的目标是开发一种创新的算法和软件工具，以减少 安全事件报告分类和分析的负担，以便将报告数据转换为 可操作的见解。使安全事件数据更具可操作性将支持安全性的主动识别 在病人受到伤害之前。我们将通过以下方式实现我们的研究目标：（1）开发 自然语言处理算法，将安全事件报告分类为可采取行动的用药错误 （2）开发原型软件，将自动分类和可视化安全 事件报告，以支持趋势识别;以及（3）与医院和 患者安全组织安全分析师。 该项目利用了研究团队在人为因素和安全科学方面的广泛专业知识， 包括计算机科学，特别是关于信息检索和数据分类。我们的研究 团队包括患者安全组织，并与计算机科学部门合作， 乔治敦大学。该提案直接与AHRQ的优先领域，使医疗保健更安全。 这项研究的贡献将包括扩大我们对自然语言处理的理解 及其在临床文本分类中的应用，可视化分析的进展，以及软件的开发 支持患者安全分析的工具。这项研究的成果将服务于两个医疗保健组织 和患者安全组织，使他们能够更有效地分析安全报告数据。

Abstract The objective of the proposed research is to develop an innovative algorithms and a software tool to reduce the burden of safety event report classification and analysis so that report data can be transformed to actionable insights. Making safety event data more actionable will support the proactive identification of safety hazards before patients are harmed. We will achieve our research objective through (1) the development of natural language processing algorithms to classify safety event reports into actionable medication error categories; (2) the development of prototype software that will automatically categorize and visualize safety event reports to support trend identification; and (3) the pilot testing of prototype software with hospital and patient safety organization safety analysts. This project utilizes the extensive expertise of the research team in human factors and safety science, including computer science, specifically regarding information retrieval and data classification. Our research team includes patient safety organizations and collaboration with the computer science department at Georgetown University. The proposal is directly aligned with AHRQ’s priority area of making health care safer. Contributions from this research will include an expansion of our understanding of natural language processing and its application to categorizing clinical text, advances in visual analytics, and the development of a software tool to support patient safety analysts. The outputs of this research will serve both healthcare organizations and patient safety organizations allowing them to more efficiently and effectively analyze safety report data.