ICU sedation, delirium & recovery
Machine learning prediction of unexpected readmission or death after discharge from intensive care: A retrospective cohort study
Authors' conclusion
ML models were able to identify patients that will not need unplanned ICU readmission and will not die within 48 h after discharge.
Study context
- Population
- Adults (≥18 years) discharged from intensive care units (ICUs) at Kepler University Hospital, Linz, Austria, 2010-01-01 to 2019-10-31
- Exposure
- Machine learning (ML) models vs Stability and Workload Index for Transfer (SWIFT) score
- Outcome
- Unplanned ICU readmission or death
- Follow-up
- 48 h after discharge
Methodological detailssingle-centre, observational, machine learning prediction
- Single-centre
- Observational
- Machine learning prediction
Main finding
As reported in the abstract
Prediction of unplanned ICU readmission or death
| Group | Area under the curve of the receiver operating characteristic (AUC-ROC) | NPV |
|---|---|---|
| ML models | 0.721 ± 0.029 | 0.990 ± 0.002 |
| SWIFT | 0.618 ± 0.011 | — not stated in the abstract |
Area under the curve of the receiver operating characteristic (AUC-ROC)
ML models0.721 ± 0.029
SWIFT0.618 ± 0.011
NPV
ML models0.990 ± 0.002
SWIFT— not stated in the abstract
— not stated in the abstract
Observational data: associations, not effects.
Further findings
Feature importance (Shapley Additive Explanations)
- Most important features: heart rate, SpO2 and arterial BP
Original publication
Machine learning prediction of unexpected readmission or death after discharge from intensive care: A retrospective cohort study. J Clin Anesth. Published 14 October 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →