-
1
Acute kidney injury (AKI) is a common complication in polytrauma patients, complicating early risk stratification due to reliance on delayed serum creatinine changes.
-
2
The study developed a machine learning model using data from the first 6 hours after ICU admission to predict AKI in adult polytrauma patients.
-
3
Logistic regression achieved the highest performance in predicting AKI, with an AUC of 0.900 and high sensitivity and negative predictive value.
-
4
SHAP analysis identified key predictors for AKI, including SOFA score, weight, magnesium levels, and vital signs.
-
5
External validation showed moderate transportability of the logistic regression model, indicating the need for further prospective validation.
