EUVIMEDOdkrijte EUVIMED
Open access

Prediction of In-Hospital Cardiac Arrest in the Intensive Care Unit: Machine Learning–Based Multimodal Approach

Vorhersage des intrahospitalen Herzstillstands auf der Intensivstation: multimodaler Machine-Learning-Ansatz

Hsin-Ying Lee iD, Po-Chih Kuo iD, Frank Qian iD, Chien-Hung Li iD, Jiun-Ruey Hu iD, Wan-Ting Hsu iD, Hong-Jie Jhou iD, Po-Huang Chen iD, Cho-Hao Lee iD, Chin-Hua Su iD, Po-Chun Liao iD, I-Ju Wu iD and 1 more

JMIR Medical Informatics · 2024-07-23 · 12 · e49142-e49142

DOI: 10.2196/49142 ↗

Auf einen Blick

Ein multimodales Machine-Learning-Modell sagte einen intrahospitalen Herzstillstand bis zu 13 Stunden im Voraus vorher. Die externe Validierung zeigte ähnliche Ergebnisse.

Fragestellung

Kann ein multimodales Machine-Learning-Modell den intrahospitalen Herzstillstand auf der Intensivstation vorhersagen?

Methode

Entwicklung mit MIMIC-IV, Validierung mit eICU-CRD und National Taiwan University Hospital; Random Forest, LSTM und SVM-Stacking auf Basis von Vitalzeichen und EMR-Daten.

Ergebnisse

23.909 Patient:innen in MIMIC-IV und 10.049 in eICU-CRD; 452 bzw. 85 mit IHCA. AUC 0,85 (95 %-KI 0,815–0,885) 13 Stunden vor dem Ereignis; extern 0,81 (95 %-KI 0,763–0,851) und 0,945 (95 %-KI 0,934–0,956).

Schlussfolgerung der Autor:innen

Mit Vitalzeichen und Informationen aus der elektronischen Krankenakte ließ sich eine klinische Verschlechterung bis zu 13 Stunden vor IHCA erkennen; das validierte Tool könnte bei der Identifikation gefährdeter Patient:innen helfen.

Kernaussagen

  • AUC 0,85 13 Stunden vor IHCA in MIMIC-IV
  • Externe Validierung mit AUC 0,81 und 0,945
  • Modell nutzt Vitalzeichen und EMR-Daten

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

At a glance

Journal article in JMIR Medical Informatics, published online on 2024-07-23. Written by Hsin-Ying Lee and 12 co-authors. Institution: Department of Medicine, College of Medicine, National Taiwan University, Taipei, Taiwan. Cited 27 times so far (Crossref); 35 references. The full text is freely available in PubMed Central.

Auf einen BlickPovzetekSubjectsBibliografski podatkiCiteRelated studies

Abstract (original language)

Background

Early identification of impending in-hospital cardiac arrest (IHCA) improves clinical outcomes but remains elusive for practicing clinicians.

Objective

We aimed to develop a multimodal machine learning algorithm based on ensemble techniques to predict the occurrence of IHCA.

Methods

Our model was developed by the Multiparameter Intelligent Monitoring of Intensive Care (MIMIC)–IV database and validated in the Electronic Intensive Care Unit Collaborative Research Database (eICU-CRD). Baseline features consisting of patient demographics, presenting illness, and comorbidities were collected to train a random forest model. Next, vital signs were extracted to train a long short-term memory model. A support vector machine algorithm then stacked the results to form the final prediction model.

Results

Of 23,909 patients in the MIMIC-IV database and 10,049 patients in the eICU-CRD database, 452 and 85 patients, respectively, had IHCA. At 13 hours in advance of an IHCA event, our algorithm had already demonstrated an area under the receiver operating characteristic curve of 0.85 (95% CI 0.815‐0.885) in the MIMIC-IV database. External validation with the eICU-CRD and National Taiwan University Hospital databases also presented satisfactory results, showing area under the receiver operating characteristic curve values of 0.81 (95% CI 0.763-0.851) and 0.945 (95% CI 0.934-0.956), respectively.

Conclusions

Using only vital signs and information available in the electronic medical record, our model demonstrates it is possible to detect a trajectory of clinical deterioration up to 13 hours in advance. This predictive tool, which has undergone external validation, could forewarn and help clinicians identify patients in need of assessment to improve their overall prognosis.

Subjects

Author keywords

cardiac arrestearly warning scoresintensive caremachine learningmedical emergency teammortality

Authors and institutions

  • Department of Medicine, College of Medicine, National Taiwan University, Taipei, Taiwan.
  • Department of Computer Science, National Tsing Hua University, Hsinchu, Taiwan.
  • Section of Cardiovascular Medicine, Boston Medical Center, Boston, MA, United States.
  • Section of Cardiovascular Medicine, Boston University Chobanian & Avedisian School of Medicine, Boston, MA, United States.
  • Department of Internal Medicine, Yale School of Medicine, New Haven, CT, United States.
  • Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, United States.

ORCID profiles: Hsin-Ying Lee, Po-Chih Kuo, Frank Qian, Chien-Hung Li, Jiun-Ruey Hu, Wan-Ting Hsu, Hong-Jie Jhou, Po-Huang Chen, Cho-Hao Lee, Chin-Hua Su, Po-Chun Liao, I-Ju Wu, Chien-Chang Lee

Cite

APA

Lee, H. Y., Kuo, P. C., Qian, F., Li, C. H., Hu, J. R., Hsu, W. T., Jhou, H. J., Chen, P. H., Lee, C. H., Su, C. H., Liao, P. C., Wu, I. J., & Lee, C. C. (2024). Prediction of In-Hospital Cardiac Arrest in the Intensive Care Unit: Machine Learning–Based Multimodal Approach. JMIR Medical Informatics, 12, e49142-e49142. https://doi.org/10.2196/49142

Vancouver

Lee HY, Kuo PC, Qian F, Li CH, Hu JR, Hsu WT, et al. Prediction of In-Hospital Cardiac Arrest in the Intensive Care Unit: Machine Learning–Based Multimodal Approach. JMIR Medical Informatics. 2024;12:e49142-e49142. doi:10.2196/49142

BibTeX
@article{Lee2024,
  author = {Lee, Hsin-Ying and Kuo, Po-Chih and Qian, Frank and Li, Chien-Hung and Hu, Jiun-Ruey and Hsu, Wan-Ting and Jhou, Hong-Jie and Chen, Po-Huang and Lee, Cho-Hao and Su, Chin-Hua and Liao, Po-Chun and Wu, I-Ju and Lee, Chien-Chang},
  title = {Prediction of In-Hospital Cardiac Arrest in the Intensive Care Unit: Machine Learning–Based Multimodal Approach},
  journal = {JMIR Medical Informatics},
  year = {2024},
  volume = {12},
  pages = {e49142--e49142},
  doi = {10.2196/49142}
}
Start for free →