RISK PREDICTION MODEL FOR ACUTE KIDNEY INJURY IN PATIENTS WITH SEVERE ACUTE PANCREATITIS.

Abstract

This research is dedicated to pinpointing the risk factors associated with acute kidney injury (AKI) in patients diagnosed with severe acute pancreatitis (SAP). Additionally, it endeavors to construct a nomogram that can accurately predict the risk of AKI in such patients. A total of 60 SAP patients, who were admitted to our hospital between July and December 2024, were selected as the research subjects. These patients were divided into a control group, which did not experience AKI, and an observation group, which had developed AKI. Extensive general information, including age, gender, and body mass index, as well as comprehensive clinical data, such as various laboratory test results and disease severity scores, were meticulously collected for all patients. Univariate analysis was initially carried out to screen out potential factors related to AKI. Subsequently, the factors that demonstrated statistical significance in the univariate analysis were incorporated into the multivariate Logistic regression analysis to identify the independent risk factors for AKI in SAP patients. By leveraging R software, a nomogram for predicting the risk of AKI in SAP patients was successfully established, with its foundation being the relevant factors determined from the univariate and multivariate analyses. The predictive performance of this nomogram was evaluated by means of the concordance index (C-index). To further validate the stability and accuracy of the model, the Bootstrap method was adopted. This involved conducting resampling with replacement 1000 times within the development cohort of the nomogram. In each resampling process, a resampled dataset with an identical sample size was constructed and utilized as the training set, while the original development cohort served as the validation set. Through this repeated process, the relatively calibrated C-index was calculated to comprehensively assess the performance of the model.