학술논문
Improving ED Emergency Severity Index Acuity Assignment Using Machine Learning and Clinical Natural Language Processing.
Document Type
Academic Journal
Author
Source
Publisher: Mosby-Year Book Country of Publication: United States NLM ID: 7605913 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1527-2966 (Electronic) Linking ISSN: 00991767 NLM ISO Abbreviation: J Emerg Nurs
Subject
Language
English
Abstract
Introduction: Triage is critical to mitigating the effect of increased volume by determining patient acuity, need for resources, and establishing acuity-based patient prioritization. The purpose of this retrospective study was to determine whether historical EHR data can be used with clinical natural language processing and machine learning algorithms (KATE) to produce accurate ESI predictive models.
Methods: The KATE triage model was developed using 166,175 patient encounters from two participating hospitals. The model was tested against a random sample of encounters that were correctly assigned an acuity by study clinicians using the Emergency Severity Index (ESI) standard as a guide.
Results: At the study sites, KATE predicted accurate ESI acuity assignments 75.7% of the time compared with nurses (59.8%) and the average of individual study clinicians (75.3%). KATE's accuracy was 26.9% higher than the average nurse accuracy (P <.001). On the boundary between ESI 2 and ESI 3 acuity assignments, which relates to the risk of decompensation, KATE's accuracy was 93.2% higher, with 80% accuracy compared with triage nurses 41.4% accuracy (P <.001).
Discussion: KATE provides a triage acuity assignment more accurate than the triage nurses in this study sample. KATE operates independently of contextual factors, unaffected by the external pressures that can cause under triage and may mitigate biases that can negatively affect triage accuracy. Future research should focus on the impact of KATE providing feedback to triage nurses in real time, on mortality and morbidity, ED throughput, resource optimization, and nursing outcomes.
(Copyright © 2020 Emergency Nurses Association. Published by Elsevier Inc. All rights reserved.)
Methods: The KATE triage model was developed using 166,175 patient encounters from two participating hospitals. The model was tested against a random sample of encounters that were correctly assigned an acuity by study clinicians using the Emergency Severity Index (ESI) standard as a guide.
Results: At the study sites, KATE predicted accurate ESI acuity assignments 75.7% of the time compared with nurses (59.8%) and the average of individual study clinicians (75.3%). KATE's accuracy was 26.9% higher than the average nurse accuracy (P <.001). On the boundary between ESI 2 and ESI 3 acuity assignments, which relates to the risk of decompensation, KATE's accuracy was 93.2% higher, with 80% accuracy compared with triage nurses 41.4% accuracy (P <.001).
Discussion: KATE provides a triage acuity assignment more accurate than the triage nurses in this study sample. KATE operates independently of contextual factors, unaffected by the external pressures that can cause under triage and may mitigate biases that can negatively affect triage accuracy. Future research should focus on the impact of KATE providing feedback to triage nurses in real time, on mortality and morbidity, ED throughput, resource optimization, and nursing outcomes.
(Copyright © 2020 Emergency Nurses Association. Published by Elsevier Inc. All rights reserved.)