EUVIMEDEUVIMED entdecken
Multizentrische StudieOpen Access

Machine Learning Algorithms to Predict Venous Thromboembolism in Patients With Sepsis in the Intensive Care Unit: Multicenter Retrospective Study

Machine-Learning-Algorithmen zur Vorhersage von Venenthromboembolien bei Patient:innen mit Sepsis auf der Intensivstation: multizentrische retrospektive Studie

Yan Zhang iD, Xia Ren iD, Luojie Liu iD, Junjie Zha iD, Yijie Gu iD, Hongwei Ye iD

JMIR Medical Informatics · 30.01.2026 · 14 · e80969

DOI: 10.2196/80969 ↗

Auf einen Blick

Ein interpretierbares ML-Modell sagte VTE bei ICU-Patient:innen mit Sepsis gut voraus. Das Light-Gradient-Boosting-Machine-Modell erzielte die beste Leistung, auch in der externen Validierung.

Fragestellung

Lässt sich mit einem interpretierbaren ML-Modell eine VTE bei ICU-Patient:innen mit Sepsis früh vorhersagen?

Methode

Multizentrische retrospektive Studie mit Entwicklung und interner Validierung in MIMIC-IV sowie externer Validierung in einem unabhängigen Kollektiv; 9 ML-Modelle wurden verglichen.

Ergebnisse

25.197 Patient:innen aus MIMIC-IV und 328 aus der externen Kohorte. VTE: 844/25.197 (3,4 %) und 30/328 (9,2 %). Bestes Modell: Light Gradient Boosting Machine, AUC 0,956 intern und 0,786 extern.

Schlussfolgerung der Autor:innen

Die Autor:innen entwickelten ein leistungsstarkes und interpretierbares ML-Modell zur Vorhersage von VTE bei ICU-Patient:innen mit Sepsis. Es zeigte Robustheit zwischen den Kohorten und bessere Leistung bei schwerer Sepsis.

Kernaussagen

  • Bestes Modell war Light Gradient Boosting Machine.
  • AUC 0,956 intern und 0,786 extern.
  • SHAP nannte zentrale Venenkatheterisierung und Laborwerte als wichtige Prädiktoren.

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

Kurz gefasst

Multizentrische Studie in der Zeitschrift JMIR Medical Informatics, online erschienen am 30.01.2026. Verfasst von Yan Zhang und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Emergency and Critical Care Medicine, Changshu No.1 People's Hospital, Changshu, Jiangsu, China. Hauptthemen laut MEDLINE: Sepsis, Venous Thromboembolism, Machine Learning. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 43 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.

Auf einen BlickAbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Venous thromboembolism (VTE) is a common and severe complication in intensive care unit (ICU) patients with sepsis. Conventional risk stratification tools lack sepsis-specific features and may inadequately capture complex, nonlinear interactions among clinical variables.

Objective

This study aimed to develop and validate an interpretable machine learning (ML) model for the early prediction of VTE in ICU patients with sepsis.

Methods

This multicenter retrospective study used data from the Medical Information Mart for Intensive Care IV database for model development and internal validation, and an independent cohort from Changshu Hospital for external validation. Candidate predictors were selected through univariate analysis, followed by least absolute shrinkage and selection operator regression. Retained variables were used in multivariable logistic regression to identify independent predictors, which were then used to develop 9 ML models, including categorical boosting, decision tree, k-nearest neighbor, light gradient boosting machine, logistic regression, multilayer perceptron, naive Bayes, random forest, and support vector machine. Performance was evaluated by discrimination (area under the curve [AUC]), calibration, and clinical use (decision curve analysis). A subgroup analysis stratified by the Sequential Organ Failure Assessment score was conducted in the external cohort to assess model stability across sepsis severity levels. Model interpretability was assessed using Shapley Additive Explanations (SHAP) to quantify the contribution of features to the predicted risk.

Results

A total of 25,197 patients from the Medical Information Mart for Intensive Care IV cohort and 328 patients from the external cohort were included, with VTE incidences of 844 out of 25,197 (3.4%) and 30 out of 328 (9.2%), respectively. The light gradient boosting machine model performed best, achieving an AUC of 0.956 in internal validation. Despite the higher VTE incidence and clinical severity in the external validation, the model maintained robust generalization with an AUC of 0.786. Notably, the model achieved enhanced discriminative ability in the severe sepsis subgroup (Sequential Organ Failure Assessment score >6) with an AUC of 0.816, compared with 0.769 in the mild to moderate sepsis subgroup. Calibration curves indicated strong agreement between predicted and observed outcomes, and decision curve analysis showed superior net benefit across clinically relevant thresholds. SHAP analysis identified central venous catheterization, serum chloride and bicarbonate levels, arterial catheterization, and prolonged partial thromboplastin time as the most influential predictors. Partial dependence plots revealed both linear and nonlinear associations between these variables and VTE risk. Individual-level force plots further enhanced interpretability by visualizing personalized risk profiles.

Conclusions

We developed a high-performing and interpretable ML model for predicting VTE in ICU patients with sepsis. The model demonstrated robustness across cohorts and enhanced performance in the severe sepsis population. By integrating diverse clinical data and leveraging SHAP for transparent explanations, this tool may support personalized prophylaxis and early diagnostic strategies.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

SepsisVenous ThromboembolismMachine LearningHumansRetrospective StudiesIntensive Care UnitsFemaleMaleMiddle AgedPredictive Learning ModelsBoosting Machine Learning AlgorithmsAgedLogistic ModelsRisk AssessmentPrediction AlgorithmsRandom ForestClassification AlgorithmsRisk FactorsAlgorithms

Stichwörter der Autor:innen

MIMIC-IVSHAPmachine learningsepsisvenous thromboembolism

Autor:innen und Institutionen

  • Department of Emergency and Critical Care Medicine, Changshu No.1 People's Hospital, Changshu, Jiangsu, China.
  • Department of Gastroenterology, Changshu No.1 People's Hospital, Changshu, Jiangsu, China.

ORCID-Profile: Yan Zhang, Xia Ren, Luojie Liu, Junjie Zha, Yijie Gu, Hongwei Ye

Zitieren

APA

Zhang, Y., Ren, X., Liu, L., Zha, J., Gu, Y., & Ye, H. (2026). Machine Learning Algorithms to Predict Venous Thromboembolism in Patients With Sepsis in the Intensive Care Unit: Multicenter Retrospective Study. JMIR Medical Informatics, 14, e80969. https://doi.org/10.2196/80969

Vancouver

Zhang Y, Ren X, Liu L, Zha J, Gu Y, Ye H. Machine Learning Algorithms to Predict Venous Thromboembolism in Patients With Sepsis in the Intensive Care Unit: Multicenter Retrospective Study. JMIR Medical Informatics. 2026;14:e80969. doi:10.2196/80969

BibTeX
@article{Zhang2026,
  author = {Zhang, Yan and Ren, Xia and Liu, Luojie and Zha, Junjie and Gu, Yijie and Ye, Hongwei},
  title = {Machine Learning Algorithms to Predict Venous Thromboembolism in Patients With Sepsis in the Intensive Care Unit: Multicenter Retrospective Study},
  journal = {JMIR Medical Informatics},
  year = {2026},
  volume = {14},
  pages = {e80969},
  doi = {10.2196/80969}
}
Kostenlos starten →