A risk nomogram for 30-day mortality in Chinese patients with acute pancreatitis using LASSO-logistic regression

This study proposed to explore the 30-day mortality risk factors in patients with acute pancreatitis (AP) and construct a prognosis nomogram based on the Least absolute shrinkage and selection operator (LASSO) logistic regression. A retrospective study on 965 adult AP patients started from January 2017 and December 2019 was conducted. Feature selection is carried out by using LASSO regression, and the model was established through logistic regression (P < 0.05). The area under the receiver operating characteristic curve (AUC), calibration curves, bootstrap and decision curve analysis (DCA) were utilized for evaluating the performance of the nomogram. A sum of 965 eligible patients were participated, of whom 922 were assigned into a survival group and 43 in a non-survival group. Six independent predictors were identified as the most valuable characteristics in AP patients, including age, activated partial thromboplastin time (APTT), direct bilirubin (DBIL), lactate dehydrogenase (LDH), total protein (TP) and blood urea nitrogen (UREA). The AUC of the nomogram was 0.862 (0.806-0.918). The DCA curve indicates that this nomogram possesses good clinical application value. The nomogram we constructed demonstrates a strong capability in predicting the 30-day mortality of AP patients.