EUVIMEDScopri EUVIMED
BeobachtungsstudieOpen access

Machine Learning–Based Prediction of Delirium and Risk Factor Identification in Intensive Care Unit Patients With Burns: Retrospective Observational Study

Machine-Learning-basierte Vorhersage von Delir und Identifikation von Risikofaktoren bei Intensivpatient:innen mit Verbrennungen: retrospektive Beobachtungsstudie

Ryo Esumi iD, Hiroki Funao iD, Eiji Kawamoto iD, Ryota Sakamoto iD, Asami Ito-Masui iD, Fumito Okuno iD, Toru Shinkai iD, Atsuya Hane iD, Kaoru Ikejiri iD, Yuichi Akama iD, Arong Gaowa iD, Eun Jeong Park iD and 3 more

JMIR Formative Research · 2025-03-05 · 9 · e65190

DOI: 10.2196/65190 ↗

Auf einen Blick

Bei 82 Intensivpatient:innen mit Verbrennungen sagte ein Machine-Learning-Modell Delir während des ICU-Aufenthalts anhand von Daten vom Aufnahmetag voraus. Die besten Modelle erreichten eine hohe AUC.

Fragestellung

Lässt sich Delir bei Intensivpatient:innen mit Verbrennungen mit Machine-Learning-Modellen aus Daten des ersten ICU-Tags vorhersagen, und welche Faktoren sind damit verbunden?

Methode

Retrospektive Beobachtungsstudie bei 82 Patient:innen mit Verbrennungen in einer ICU; 70 Variablen bei Aufnahme; Delir-Screening alle 8 Stunden; 10 Machine-Learning-Methoden; Shapley-Analyse für Risikofaktoren.

Ergebnisse

Logistische Regression: mittlere AUC 0,906 (SD 0,073); SVM 0,897 (0,056); kNN 0,894 (0,060); neuronales Netz 0,857 (0,058); Random Forest 0,850 (0,074). Wichtige Faktoren: 24-h-Urinmenge, Sauerstoffsättigung, Verbrennungsfläche, Gesamtbilirubin, Intubation bei Aufnahme.

Schlussfolgerung der Autor:innen

Die Studie zeigte, dass mit Vitalparametern und Blutdaten bei ICU-Aufnahme trainierte Machine-Learning-Modelle Delir bei Verbrennungspatient:innen während des ICU-Aufenthalts vorhersagen können.

Kernaussagen

  • 82 Patient:innen mit Verbrennungen auf der ICU
  • Beste AUC: logistische Regression 0,906
  • Wichtige Faktoren: 24-h-Urinmenge, SpO2, Verbrennungsfläche

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

Beobachtungsstudie in JMIR Formative Research, published online on 2025-03-05. Written by Ryo Esumi and 14 co-authors. Institution: Department of Molecular Pathobiology and Cell Adhesion Biology, Mie University Graduate School of Medicine, Mie University, Tsu, Japan. Major MEDLINE topics: Delirium, Machine Learning, Burns. Cited 9 times so far (Crossref); 34 references. The full text is freely available in PubMed Central.

Auf einen BlickAbstractSubjectsDati bibliograficiCiteRelated studies

Abstract (original language)

Background

The incidence of delirium in patients with burns receiving treatment in the intensive care unit (ICU) is high, reaching up to 77%, and has been associated with increased mortality rates. Therefore, early identification of patients at high risk of delirium onset is essential for improving treatment strategies.

Objective

This study aimed to create a machine learning model for predicting delirium in patients with burns during their ICU stay using patient data from the first day of ICU admission and identify predictive factors for ICU delirium in patients with burns.

Methods

This study focused on 82 patients with burns aged ≥18 years who were admitted to the ICU at Mie University Hospital for ≥24 hours between January 2015 and June 2023. In total, 70 variables were measured in patients upon ICU admission and used as explanatory variables in the ICU delirium prediction model. Delirium was assessed using the Intensive Care Delirium Screening Checklist every 8 hours after ICU admission. A total of 10 different machine learning methods were used to predict ICU delirium. Multiple receiver operating characteristic curves were plotted for various machine learning models, and the area under the curve (AUC) for each was compared. In addition, the top 15 risk factors contributing to delirium onset were identified using Shapley additive explanations analysis.

Results

Among the 10 machine learning models tested, logistic regression (mean AUC 0.906, SD 0.073), support vector machine (mean AUC 0.897, SD 0.056), k-nearest neighbor (mean AUC 0.894, SD 0.060), neural network (mean AUC 0.857, SD 0.058), random forest (mean AUC 0.850, SD 0.074), adaptive boosting (mean AUC 0.832, SD 0.094), gradient boosting machine (mean AUC 0.821, SD 0.074), and naïve Bayes (mean AUC 0.827, SD 0.095) demonstrated the highest accuracy in predicting ICU delirium. Specifically, 24-hour urine output (from ICU admission to 24 hours), oxygen saturation, burn area, total bilirubin level, and intubation upon ICU admission were identified as the major risk factors for delirium onset. In addition, variables, such as the proportion of white blood cell fractions, including monocytes; methemoglobin concentration; and respiratory rate, were identified as important risk factors for ICU delirium.

Conclusions

This study demonstrated the ability of machine learning models trained using vital signs and blood data upon ICU admission to predict delirium in patients with burns during their ICU stay.

Subjects

MeSH terms (MEDLINE)

DeliriumMachine LearningBurnsHumansMaleFemaleIntensive Care UnitsRetrospective StudiesRisk FactorsMiddle AgedAdultAgedRisk AssessmentROC Curve

Author keywords

AIartificial intelligenceburnsdeliriumintensive care unitmachine learningprediction model

Authors and institutions

  • Department of Molecular Pathobiology and Cell Adhesion Biology, Mie University Graduate School of Medicine, Mie University, Tsu, Japan.
  • Department of Practical Nursing, Mie University Graduate School of Medicine, Tsu city, Japan.
  • Department of Medical Informatics, Mie University Hospital, Tsu city, Japan.
  • Department of Emergency and Critical Care Center, Mie University Hospital, Tsu city, Japan.
  • Department of Rehabilitation Medicine, Mie University Graduate School of Medicine, Tsu city, Japan.
  • Department of Anesthesiology, Mie University Hospital, Tsu city, Japan.

ORCID profiles: Ryo Esumi, Hiroki Funao, Eiji Kawamoto, Ryota Sakamoto, Asami Ito-Masui, Fumito Okuno, Toru Shinkai, Atsuya Hane, Kaoru Ikejiri, Yuichi Akama, Arong Gaowa, Eun Jeong Park, Ryo Momosaki, Ryuji Kaku, Motomu Shimaoka

Cite

APA

Esumi, R., Funao, H., Kawamoto, E., Sakamoto, R., Ito-Masui, A., Okuno, F., Shinkai, T., Hane, A., Ikejiri, K., Akama, Y., Gaowa, A., Park, E. J., Momosaki, R., Kaku, R., & Shimaoka, M. (2025). Machine Learning–Based Prediction of Delirium and Risk Factor Identification in Intensive Care Unit Patients With Burns: Retrospective Observational Study. JMIR Formative Research, 9, e65190. https://doi.org/10.2196/65190

Vancouver

Esumi R, Funao H, Kawamoto E, Sakamoto R, Ito-Masui A, Okuno F, et al. Machine Learning–Based Prediction of Delirium and Risk Factor Identification in Intensive Care Unit Patients With Burns: Retrospective Observational Study. JMIR Formative Research. 2025;9:e65190. doi:10.2196/65190

BibTeX
@article{Esumi2025,
  author = {Esumi, Ryo and Funao, Hiroki and Kawamoto, Eiji and Sakamoto, Ryota and Ito-Masui, Asami and Okuno, Fumito and Shinkai, Toru and Hane, Atsuya and Ikejiri, Kaoru and Akama, Yuichi and Gaowa, Arong and Park, Eun Jeong and Momosaki, Ryo and Kaku, Ryuji and Shimaoka, Motomu},
  title = {Machine Learning–Based Prediction of Delirium and Risk Factor Identification in Intensive Care Unit Patients With Burns: Retrospective Observational Study},
  journal = {JMIR Formative Research},
  year = {2025},
  volume = {9},
  pages = {e65190},
  doi = {10.2196/65190}
}
Start for free →