376 Suitability of clinical workflows for automation
376 Suitability of clinical workflows for automation
复制标题
376 临床工作流程自动化的适用性
DOI：
10.1016/j.jid.2021.02.398
复制
发表时间：
2021
影响因子：
6.5
通讯作者：
Pentland, B.
中科院分区：
文献类型：
--
作者：
Wolf, J. Ryan;Kim, I.;Xie, Y.;Pentland, A.;Pentland, B.
人们对使用信息技术支持医疗保健工作流越来越感兴趣；然而，并不是所有的工作流都适合自动化。我们开发了两个指标，制定的复杂性(EC)和上下文依赖(CD)，以识别和排名临床工作流的自动化适应性。EC是工作流中路径数量的函数；路径越多，复杂性越大。CD是工作流程受上下文细节影响程度的函数。高CD表明工作流程在很大程度上取决于何时、何地和由谁执行。EC和CD指示工作流可能难以映射、监视和控制，这表明该工作流不太适合自动化。在这项研究中，我们使用电子病历(EMR)中来自24个不同门诊诊所(皮肤科、骨科和儿科肿瘤科)的143,347次就诊的临床文档数据计算EC和CD。手术病理数据被纳入为一个简单的工作流程比较器。我们使用EC和CD对诊所的自动化适宜性进行排序。EC和CD之间有很强的相关性(Spearman r=0.55，p<；0.05)。外科病理学工作流由几条路径组成，并且几乎与上下文无关。相比之下，皮肤科诊所有超过167,000条路径，整形外科诊所有数百万条路径。皮肤科和整形外科的工作流程都高度依赖于背景。尽管皮肤科诊所极其复杂，但它们似乎比其他门诊诊所更容易实现自动化。我们的结论是，EC和CD这两个指标可以识别适合自动化并可能受益于自动化的医疗工作流。本研究得到了罗切斯特大学国家自然科学基金(SES-1734237)的资助(UL1TR002001)。
There is increasing interest in using information technology to support healthcare workflows; however not all workflows are suitable for automation. We developed two metrics, enacted complexity (EC) and contextual dependence (CD), to identify and rank clinic workflows on their amenability for automation. EC is a function of the number of paths in a workflow; more paths indicate greater complexity. CD is a function of how much a workflow is influenced by contextual specifics. High CD indicates that a workflow greatly depends on when, where, and by whom it is performed. EC and CD are indicators that a workflow may be difficult to map, monitor, and control, suggesting that the workflow is a less feasible target for automation. In this study, we computed EC and CD using clinical documentation data from the electronic medical record (EMR) for 143,347 visits from 24 different outpatient clinics (Dermatology, Orthopedic Surgery, and Pediatric Oncology). Surgical Pathology data were included as a simple workflow comparator. We used EC and CD to rank order the clinics for automation suitability. EC and CD showed strong correlation (Spearman r= 0.55, p< 0.05). Surgical Pathology workflow consisted of a handful of paths and is very nearly context independent. In contrast, Dermatology clinics had over 167,000 paths and Orthopedic Surgery clinic had millions of paths. Both Dermatology and Orthopedic Surgery workflows were highly context dependent. Although Dermatology clinics were extremely complex, they appeared more amenable to automation than the other outpatient clinics. We conclude that the two metrics, EC and CD, can identify healthcare workflows that are suitable for and may benefit from automation. This research was supported by NSF (SES-1734237), University of Rochester CTSA (UL1 TR002001).