Key result
Machine learning outperforms the SWIFT score in predicting 48-hour ICU readmission or death.
Why the study?
Discharge from the ICU requires evaluating stability, as organ dysfunction or instability following discharge presents potentially life-threatening situations for patients.
Do machine learning models improve the prediction of unplanned ICU readmission or death compared to the SWIFT score in adult ICU patients?
Cohort
No
Do machine learning models improve the prediction of unplanned ICU readmission or death compared to the SWIFT score in adult ICU patients?
Effect estimate: AUC-ROC 0.721
Machine learning models outperformed the SWIFT score in predicting which patients will not require unplanned ICU readmission or die within 48 hours of discharge.
ML models may aid post-ICU risk stratification; observational data leave clinical adoption open pending prospective validation.
BACKGROUND: Intensive care units (ICUs) harbor the sickest patients with the utmost needs of medical care. Discharge from ICU needs to consider the reason for admission and stability after ICU care. Organ dysfunction or instability after ICU discharge constitute potentially life-threatening situations for patients. METHODS: This is a single center, observational, retrospective cohort study conducted at ICUs at the Kepler University Hospital in Linz, Austria. Patients aged 18 years and above admitted to the study center's ICUs between 2010-01-01 and 2019-10-31 were included in the study. Patients transferred to another ICU, discharged to a different hospital or home, or that died during their ICU stay were excluded. We used machine learning (ML) models to predict unplanned ICU readmission or death using an internal dataset or MIMIC-IV as training data and compared the models with the Stability and Workload Index for Transfer (SWIFT) score. Further, we evaluated the influence of features on the models using Shapley Additive Explanations. RESULTS: The best ML models achieved an area under the curve of the receiver operating characteristic (AUC-ROC) of 0.721 ± 0.029 and a high negative predictive value (NPV) of 0.990 ± 0.002. The most important features were heart rate, peripheral oxygen saturation and arterial blood pressure. Performance of the SWIFT score was worse than the ML models (best AUC-ROC 0.618 ± 0.011). CONCLUSIONS: ML models were able to identify patients that will not need unplanned ICU readmission and will not die within 48 h after discharge.
No takes yet. Share an insight, caveat, or question.
Tschoellitsch et al. (2024) conducted a cohort in Intensive care unit discharge. Machine learning models vs. SWIFT score was evaluated on Unplanned ICU readmission or death (AUC-ROC 0.721). Machine learning models outperformed the SWIFT score in predicting unplanned ICU readmission or death within 48 hours after discharge (AUC-ROC 0.721 vs 0.618).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: