Machine learning for prediction of septic shock at initial triage in emergency department

Joonghee Kim, Hyung Lan Chang, Doyun Kim, Dong Hyun Jang, Inwon Park, Kyuseok Kim

Research output: Contribution to journalArticle

Abstract

Background: We hypothesized utilizing machine learning (ML) algorithms for screening septic shock in ED would provide better accuracy than qSOFA or MEWS. Methods: The study population was adult (≥20 years) patients visiting ED for suspected infection. Target event was septic shock within 24 h after arrival. Demographics, vital signs, level of consciousness, chief complaints (CC) and initial blood test results were used as predictors. CC were embedded into 16-dimensional vector space using singular value decomposition. Six base learners including support vector machine, gradient-boosting machine, random forest, multivariate adaptive regression splines and least absolute shrinkage and selection operator and ridge regression and their ensembles were tested. We also trained and tested MLP networks with various setting. Results: A total of 49,560 patients were included and 4817 (9.7%) had septic shock within 24 h. All ML classifiers significantly outperformed qSOFA score, MEWS and their age-sex adjusted versions with their AUROC ranging from 0.883 to 0.929. The ensembles of the base classifiers showed the best performance and addition of CC embedding was associated with statistically significant increases in performance. Conclusions: ML classifiers significantly outperforms clinical scores in screening septic shock at ED triage.

Original languageEnglish
Pages (from-to)163-170
Number of pages8
JournalJournal of Critical Care
Volume55
DOIs
StatePublished - Feb 2020

Fingerprint

Triage
Septic Shock
Hospital Emergency Service
Vital Signs
Hematologic Tests
Consciousness
Demography
Machine Learning
Infection
Population

Keywords

  • Clinical decision support tool
  • Diagnosis
  • Emergency department triage tool
  • Machine learning
  • Prediction
  • Sepsis
  • Septic shock

Cite this

Kim, Joonghee ; Chang, Hyung Lan ; Kim, Doyun ; Jang, Dong Hyun ; Park, Inwon ; Kim, Kyuseok. / Machine learning for prediction of septic shock at initial triage in emergency department. In: Journal of Critical Care. 2020 ; Vol. 55. pp. 163-170.
@article{221c5b49ebe944e4b8c7204f771076e7,
title = "Machine learning for prediction of septic shock at initial triage in emergency department",
abstract = "Background: We hypothesized utilizing machine learning (ML) algorithms for screening septic shock in ED would provide better accuracy than qSOFA or MEWS. Methods: The study population was adult (≥20 years) patients visiting ED for suspected infection. Target event was septic shock within 24 h after arrival. Demographics, vital signs, level of consciousness, chief complaints (CC) and initial blood test results were used as predictors. CC were embedded into 16-dimensional vector space using singular value decomposition. Six base learners including support vector machine, gradient-boosting machine, random forest, multivariate adaptive regression splines and least absolute shrinkage and selection operator and ridge regression and their ensembles were tested. We also trained and tested MLP networks with various setting. Results: A total of 49,560 patients were included and 4817 (9.7{\%}) had septic shock within 24 h. All ML classifiers significantly outperformed qSOFA score, MEWS and their age-sex adjusted versions with their AUROC ranging from 0.883 to 0.929. The ensembles of the base classifiers showed the best performance and addition of CC embedding was associated with statistically significant increases in performance. Conclusions: ML classifiers significantly outperforms clinical scores in screening septic shock at ED triage.",
keywords = "Clinical decision support tool, Diagnosis, Emergency department triage tool, Machine learning, Prediction, Sepsis, Septic shock",
author = "Joonghee Kim and Chang, {Hyung Lan} and Doyun Kim and Jang, {Dong Hyun} and Inwon Park and Kyuseok Kim",
year = "2020",
month = "2",
doi = "10.1016/j.jcrc.2019.09.024",
language = "English",
volume = "55",
pages = "163--170",
journal = "Journal of Critical Care",
issn = "0883-9441",
publisher = "Elsevier BV",

}

Machine learning for prediction of septic shock at initial triage in emergency department. / Kim, Joonghee; Chang, Hyung Lan; Kim, Doyun; Jang, Dong Hyun; Park, Inwon; Kim, Kyuseok.

In: Journal of Critical Care, Vol. 55, 02.2020, p. 163-170.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Machine learning for prediction of septic shock at initial triage in emergency department

AU - Kim, Joonghee

AU - Chang, Hyung Lan

AU - Kim, Doyun

AU - Jang, Dong Hyun

AU - Park, Inwon

AU - Kim, Kyuseok

PY - 2020/2

Y1 - 2020/2

N2 - Background: We hypothesized utilizing machine learning (ML) algorithms for screening septic shock in ED would provide better accuracy than qSOFA or MEWS. Methods: The study population was adult (≥20 years) patients visiting ED for suspected infection. Target event was septic shock within 24 h after arrival. Demographics, vital signs, level of consciousness, chief complaints (CC) and initial blood test results were used as predictors. CC were embedded into 16-dimensional vector space using singular value decomposition. Six base learners including support vector machine, gradient-boosting machine, random forest, multivariate adaptive regression splines and least absolute shrinkage and selection operator and ridge regression and their ensembles were tested. We also trained and tested MLP networks with various setting. Results: A total of 49,560 patients were included and 4817 (9.7%) had septic shock within 24 h. All ML classifiers significantly outperformed qSOFA score, MEWS and their age-sex adjusted versions with their AUROC ranging from 0.883 to 0.929. The ensembles of the base classifiers showed the best performance and addition of CC embedding was associated with statistically significant increases in performance. Conclusions: ML classifiers significantly outperforms clinical scores in screening septic shock at ED triage.

AB - Background: We hypothesized utilizing machine learning (ML) algorithms for screening septic shock in ED would provide better accuracy than qSOFA or MEWS. Methods: The study population was adult (≥20 years) patients visiting ED for suspected infection. Target event was septic shock within 24 h after arrival. Demographics, vital signs, level of consciousness, chief complaints (CC) and initial blood test results were used as predictors. CC were embedded into 16-dimensional vector space using singular value decomposition. Six base learners including support vector machine, gradient-boosting machine, random forest, multivariate adaptive regression splines and least absolute shrinkage and selection operator and ridge regression and their ensembles were tested. We also trained and tested MLP networks with various setting. Results: A total of 49,560 patients were included and 4817 (9.7%) had septic shock within 24 h. All ML classifiers significantly outperformed qSOFA score, MEWS and their age-sex adjusted versions with their AUROC ranging from 0.883 to 0.929. The ensembles of the base classifiers showed the best performance and addition of CC embedding was associated with statistically significant increases in performance. Conclusions: ML classifiers significantly outperforms clinical scores in screening septic shock at ED triage.

KW - Clinical decision support tool

KW - Diagnosis

KW - Emergency department triage tool

KW - Machine learning

KW - Prediction

KW - Sepsis

KW - Septic shock

UR - http://www.scopus.com/inward/record.url?scp=85074952430&partnerID=8YFLogxK

U2 - 10.1016/j.jcrc.2019.09.024

DO - 10.1016/j.jcrc.2019.09.024

M3 - Article

AN - SCOPUS:85074952430

VL - 55

SP - 163

EP - 170

JO - Journal of Critical Care

JF - Journal of Critical Care

SN - 0883-9441

ER -