Thomas Tschoellitsch, Philipp Seidl, Carl Böck, Alexander Maletzky, Philipp Moser, Stefan Thumfart, Michael Giretzlehner, Sepp Hochreiter, Jens Meier
{"title":"Using emergency department triage for machine learning-based admission and mortality prediction.","authors":"Thomas Tschoellitsch, Philipp Seidl, Carl Böck, Alexander Maletzky, Philipp Moser, Stefan Thumfart, Michael Giretzlehner, Sepp Hochreiter, Jens Meier","doi":"10.1097/MEJ.0000000000001068","DOIUrl":null,"url":null,"abstract":"<p><strong>Aims: </strong>Patient admission is a decision relying on sparsely available data. This study aims to provide prediction models for discharge versus admission for ward observation or intensive care, and 30 day-mortality for patients triaged with the Manchester Triage System.</p><p><strong>Methods: </strong>This is a single-centre, observational, retrospective cohort study from data within ten minutes of patient presentation at the interdisciplinary emergency department of the Kepler University Hospital, Linz, Austria. We trained machine learning models including Random Forests and Neural Networks individually to predict discharge versus ward observation or intensive care admission, and 30 day-mortality. For analysis of the features' relevance, we used permutation feature importance.</p><p><strong>Results: </strong>A total of 58323 adult patients between 1 December 2015 and 31 August 2020 were included. Neural Networks and Random Forests predicted admission to ward observation with an AUC-ROC of 0.842 ± 0.00 with the most important features being age and chief complaint. For admission to intensive care, the models had an AUC-ROC of 0.819 ± 0.002 with the most important features being the Manchester Triage category and heart rate, and for the outcome 30 day-mortality an AUC-ROC of 0.925 ± 0.001. The most important features for the prediction of 30 day-mortality were age and general ward admission.</p><p><strong>Conclusion: </strong>Machine learning can provide prediction on discharge versus admission to general wards and intensive care and inform about risk on 30 day-mortality for patients in the emergency department.</p>","PeriodicalId":11893,"journal":{"name":"European Journal of Emergency Medicine","volume":null,"pages":null},"PeriodicalIF":3.1000,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"European Journal of Emergency Medicine","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1097/MEJ.0000000000001068","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2023/8/14 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"EMERGENCY MEDICINE","Score":null,"Total":0}
引用次数: 0
Abstract
Aims: Patient admission is a decision relying on sparsely available data. This study aims to provide prediction models for discharge versus admission for ward observation or intensive care, and 30 day-mortality for patients triaged with the Manchester Triage System.
Methods: This is a single-centre, observational, retrospective cohort study from data within ten minutes of patient presentation at the interdisciplinary emergency department of the Kepler University Hospital, Linz, Austria. We trained machine learning models including Random Forests and Neural Networks individually to predict discharge versus ward observation or intensive care admission, and 30 day-mortality. For analysis of the features' relevance, we used permutation feature importance.
Results: A total of 58323 adult patients between 1 December 2015 and 31 August 2020 were included. Neural Networks and Random Forests predicted admission to ward observation with an AUC-ROC of 0.842 ± 0.00 with the most important features being age and chief complaint. For admission to intensive care, the models had an AUC-ROC of 0.819 ± 0.002 with the most important features being the Manchester Triage category and heart rate, and for the outcome 30 day-mortality an AUC-ROC of 0.925 ± 0.001. The most important features for the prediction of 30 day-mortality were age and general ward admission.
Conclusion: Machine learning can provide prediction on discharge versus admission to general wards and intensive care and inform about risk on 30 day-mortality for patients in the emergency department.
期刊介绍:
The European Journal of Emergency Medicine is the official journal of the European Society for Emergency Medicine. It is devoted to serving the European emergency medicine community and to promoting European standards of training, diagnosis and care in this rapidly growing field.
Published bimonthly, the Journal offers original papers on all aspects of acute injury and sudden illness, including: emergency medicine, anaesthesiology, cardiology, disaster medicine, intensive care, internal medicine, orthopaedics, paediatrics, toxicology and trauma care. It addresses issues on the organization of emergency services in hospitals and in the community and examines postgraduate training from European and global perspectives. The Journal also publishes papers focusing on the different models of emergency healthcare delivery in Europe and beyond. With a multidisciplinary approach, the European Journal of Emergency Medicine publishes scientific research, topical reviews, news of meetings and events of interest to the emergency medicine community.
Submitted articles undergo a preliminary review by the editor. Some articles may be returned to authors without further consideration. Those being considered for publication will undergo further assessment and peer-review by the editors and those invited to do so from a reviewer pool.