Department of Health Information Technology, School of Management and Medical Informatics, Tabriz University of Medical Sciences, Tabriz, Iran.
Road Traffic Injury Research Center, Department of Emergency Medicine, Tabriz University of Medical Sciences, Tabriz, Iran.
Int J Med Inform. 2018 Jun;114:35-44. doi: 10.1016/j.ijmedinf.2018.03.008. Epub 2018 Mar 20.
Fast and accurate patient triage for the response process is a critical first step in emergency situations. This process is often performed using a paper-based mode, which intensifies workload and difficulty, wastes time, and is at risk of human errors. This study aims to design and evaluate a decision support system (DSS) to determine the triage level.
A combination of the Rule-Based Reasoning (RBR) and Fuzzy Logic Classifier (FLC) approaches were used to predict the triage level of patients according to the triage specialist's opinions and Emergency Severity Index (ESI) guidelines. RBR was applied for modeling the first to fourth decision points of the ESI algorithm. The data relating to vital signs were used as input variables and modeled using fuzzy logic. Narrative knowledge was converted to If-Then rules using XML. The extracted rules were then used to create the rule-based engine and predict the triage levels.
Fourteen RBR and 27 fuzzy rules were extracted and used in the rule-based engine. The performance of the system was evaluated using three methods with real triage data. The accuracy of the clinical decision support systems (CDSSs; in the test data) was 99.44%. The evaluation of the error rate revealed that, when using the traditional method, 13.4% of the patients were miss-triaged, which is statically significant. The completeness of the documentation also improved from 76.72% to 98.5%.
Designed system was effective in determining the triage level of patients and it proved helpful for nurses as they made decisions, generated nursing diagnoses based on triage guidelines. The hybrid approach can reduce triage misdiagnosis in a highly accurate manner and improve the triage outcomes.
在紧急情况下，快速准确地对患者进行分诊是响应过程中的关键第一步。这个过程通常采用基于纸张的模式，这会增加工作量和难度，浪费时间，并存在人为错误的风险。本研究旨在设计和评估一个决策支持系统（DSS）来确定分诊级别。
结合基于规则的推理（RBR）和模糊逻辑分类器（FLC）方法，根据分诊专家的意见和紧急严重程度指数（ESI）指南预测患者的分诊级别。RBR 用于对 ESI 算法的前四个决策点进行建模。与生命体征相关的数据被用作输入变量，并使用模糊逻辑进行建模。叙事知识被转换为 XML 的 If-Then 规则。提取的规则随后用于创建基于规则的引擎并预测分诊级别。
提取了 14 条 RBR 和 27 条模糊规则，并在基于规则的引擎中使用。使用真实分诊数据评估了系统的三种性能。临床决策支持系统（CDSS）的准确率（在测试数据中）为 99.44%。错误率的评估表明，使用传统方法时，13.4%的患者分诊错误，这具有统计学意义。文档的完整性也从 76.72%提高到 98.5%。
设计的系统在确定患者的分诊级别方面是有效的，它有助于护士在做出决策时，根据分诊指南生成护理诊断。混合方法可以以非常高的准确度减少分诊误诊，并改善分诊结果。
