ICU sedation, delirium & recovery
Interpretable machine learning models for predicting venous thromboembolism in the intensive care unit: an analysis based on data from 207 centers
Authors' conclusion
ML models can be a reliable tool for predicting VTE in critically ill patients.
Study context
- Population
- Critically ill ICU patients, eICU Collaborative Research Database1,647 (1.5%) had venous thromboembolism (VTE)
- Participants
- 109,044 patients
- Exposure
- Random forest vs XGBoost vs support vector machine models
- Outcome
- VTE during ICU hospitalization (model prediction performance)
Methodological detailsretrospective database analysis, machine learning, fivefold cross-validation, multicentre database (eICU Collaborative Research Database version 2.0)
- Retrospective database analysis
- Machine learning
- Fivefold cross-validation
- Multicentre database (eICU Collaborative Research Database version 2.0)
Main finding
As reported in the abstract
Model performance for predicting VTE
| Group | AUC | Accuracy | Kappa | Precision | F1 score | Sensitivity | Specificity |
|---|---|---|---|---|---|---|---|
| Random Forest | 0.9378 | 0.9958 | 0.8371 | 0.9095 | 0.8393 | 0.7791 | 0.9989 |
Random Forest
AUC0.9378
Accuracy0.9958
Kappa0.8371
Precision0.9095
F1 score0.8393
Sensitivity0.7791
Specificity0.9989
Observational data: associations, not effects.
Original publication
Interpretable machine learning models for predicting venous thromboembolism in the intensive care unit: an analysis based on data from 207 centers. Crit Care. Published 24 October 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →