Development and validation of machine learning models for mortality risk stratification in acute pancreatitis...

Acute kidney injury (AKI) is a serious complication of acute pancreatitis (AP), associated with significantly increased mortality. Early identification of high-risk AP patients complicated by AKI is critical. This study aimed to develop and validate a prognostic model for early mortality prediction in this population.

Data were obtained from the Medical Information Mart for Intensive Care IV (MIMIC IV) database. Patients from 2008 to 2016 were used as a training cohort; those from 2017 to 2019 served as a validation cohort. Least absolute shrinkage and selection operator (LASSO) regression was used to identify key prognostic factors. Six models were developed and compared.

Ten predictors were selected. The LASSO-LR model showed the best discrimination, with area under the receiver operating characteristic curve (AUROC) of 0.914 in the training cohort and 0.815 in the validation cohort. Calibration slope and Brier score were 1.000 and 0.084 in training, and 0.601 and 0.137 in validation. Decision curve analysis showed superior net clinical benefit.

The LASSO-LR nomogram is a reliable and clinically useful tool for predicting mortality in patients with AP complicated by AKI.