EUVIMEDOpdag EUVIMED
Open access

Predicting Agitation-Sedation Levels in Intensive Care Unit Patients: Development of an Ensemble Model

Vorhersage von Agitations- und Sedierungsniveaus bei Intensivpatient:innen: Entwicklung eines Ensemble-Modells

Pei-Yu Dai iD, Pei-Yi Lin iD, Ruey-Kai Sheu iD, Shu-Fang Liu iD, Yu-Cheng Wu iD, Chieh-Liang Wu iD, Wei-Lin Chen iD, Chien-Chung Huang iD, Guan-Yin Lin iD, Lun-Chi Chen iD

JMIR Medical Informatics · 2025-02-26 · 13 · e63601-e63601

DOI: 10.2196/63601 ↗

Auf einen Blick

Eine machine-learning-basierte Klassifikation von Agitations- und Sedierungsniveaus wurde mit ICU-Daten entwickelt. Das Ensemble-Modell verbesserte die Sensitivität für Agitation bei hoher AUC.

Fragestellung

Lässt sich das Ausmaß von Agitation und Sedierung bei Intensivpatient:innen mit einem Ensemble-Modell automatisch klassifizieren?

Methode

Entwicklung eines Ensemble-Learning-Modells mit Daten aus der ICU-Datenbank des Taichung Veterans General Hospital (2020); Vergleich verschiedener Modellsequenzen; SHAP für die Erklärung.

Ergebnisse

Mit 20 Merkmalen und 121.303 Datenpunkten erreichte das Random-Forest-Modell AUC-Werte von 0,97 für Sedierung und 0,88 für Agitation. Das Ensemble-Modell steigerte die Sensitivität für Agitation auf 0,82; alle AUCs lagen über 0,82.

Schlussfolgerung der Autor:innen

Die Studie schlägt eine automatisierte ICU-Bewertung von Agitation und Sedierung mittels Machine Learning vor. Ensemble Learning verbessert die Sensitivität für Agitation bei hoher Genauigkeit.

Kernaussagen

  • AUC: 0,97 für Sedierung und 0,88 für Agitation
  • Ensemble-Modell: Sensitivität für Agitation 0,82
  • SHAP-Erklärungen stimmten mit klinischer Erfahrung überein

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 2025-02-26. Written by Pei-Yu Dai and 9 co-authors. Institution: Department of Digital Medicine, Taichung Veterans General Hospital, Taichung, Taiwan. Major MEDLINE topics: Intensive Care Units, Psychomotor Agitation, Machine Learning, Hypnotics and Sedatives. Cited 4 times so far (Crossref); 26 references. The full text is freely available in PubMed Central.

Auf einen BlickAbstractSubjectsBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Background

Agitation and sedation management is critical in intensive care as it affects patient safety. Traditional nursing assessments suffer from low frequency and subjectivity. Automating these assessments can boost intensive care unit (ICU) efficiency, treatment capacity, and patient safety.

Objectives

The aim of this study was to develop a machine-learning based assessment of agitation and sedation.

Methods

Using data from the Taichung Veterans General Hospital ICU database (2020), an ensemble learning model was developed for classifying the levels of agitation and sedation. Different ensemble learning model sequences were compared. In addition, an interpretable artificial intelligence approach, SHAP (Shapley additive explanations), was employed for explanatory analysis.

Results

With 20 features and 121,303 data points, the random forest model achieved high area under the curve values across all models (sedation classification: 0.97; agitation classification: 0.88). The ensemble learning model enhanced agitation sensitivity (0.82) while maintaining high AUC values across all categories (all >0.82). The model explanations aligned with clinical experience.

Conclusions

This study proposes an ICU agitation-sedation assessment automation using machine learning, enhancing efficiency and safety. Ensemble learning improves agitation sensitivity while maintaining accuracy. Real-time monitoring and future digital integration have the potential for advancements in intensive care.

Subjects

MeSH terms (MEDLINE)

Intensive Care UnitsPsychomotor AgitationMachine LearningHypnotics and SedativesHumansMaleFemaleMiddle AgedAged

Author keywords

AIICUICU careMLagitationartificial intelligenceautomationefficiencyensemble learningensemble modelexplanatory analysisintensive care unitslearning modelmachine learningpatient safety

Substances

Hypnotics and Sedatives

Authors and institutions

  • Department of Digital Medicine, Taichung Veterans General Hospital, Taichung, Taiwan.
  • Department of Nursing, Taichung Veterans General Hospital, 1650 Taiwan Boulevard Sect. 4, Taichung, 40705, Taiwan.
  • Department of Computer Science, Tunghai University, Taichung, Taiwan.
  • Department of Critical Care Medicine, Taichung Veterans General Hospital, No 1650, Section 4, Taiwan Boulevard, Xitan District, Taichung City, 407219, Taiwan, 886-04-23592525 #2002.
  • Department of Post-Baccalaureate Medicine, College of Medicine, National Chung Hsing University, Taichung, Taiwan.
  • Computer & Communications Center, Taichung Veterans General Hospital, Taichung, Taiwan.

ORCID profiles: Pei-Yu Dai, Pei-Yi Lin, Ruey-Kai Sheu, Shu-Fang Liu, Yu-Cheng Wu, Chieh-Liang Wu, Wei-Lin Chen, Chien-Chung Huang, Guan-Yin Lin, Lun-Chi Chen

Cite

APA

Dai, P. Y., Lin, P. Y., Sheu, R. K., Liu, S. F., Wu, Y. C., Wu, C. L., Chen, W. L., Huang, C. C., Lin, G. Y., & Chen, L. C. (2025). Predicting Agitation-Sedation Levels in Intensive Care Unit Patients: Development of an Ensemble Model. JMIR Medical Informatics, 13, e63601-e63601. https://doi.org/10.2196/63601

Vancouver

Dai PY, Lin PY, Sheu RK, Liu SF, Wu YC, Wu CL, et al. Predicting Agitation-Sedation Levels in Intensive Care Unit Patients: Development of an Ensemble Model. JMIR Medical Informatics. 2025;13:e63601-e63601. doi:10.2196/63601

BibTeX
@article{Dai2025,
  author = {Dai, Pei-Yu and Lin, Pei-Yi and Sheu, Ruey-Kai and Liu, Shu-Fang and Wu, Yu-Cheng and Wu, Chieh-Liang and Chen, Wei-Lin and Huang, Chien-Chung and Lin, Guan-Yin and Chen, Lun-Chi},
  title = {Predicting Agitation-Sedation Levels in Intensive Care Unit Patients: Development of an Ensemble Model},
  journal = {JMIR Medical Informatics},
  year = {2025},
  volume = {13},
  pages = {e63601--e63601},
  doi = {10.2196/63601}
}
Start for free →