A nomogram model for predicting acute kidney injury in critically ill patients with upper gastrointestinal bleeding:...

Acute kidney injury (AKI) represents a prevalent and critical complication of upper gastrointestinal bleeding (UGIB). Predicting AKI is essential for timely intervention.

This retrospective research leveraged data from the MIMIC-IV 3.1 database and data from patients hospitalized at the Second Hospital of Hebei Medical University (HMUSH). The Boruta algorithm was employed in the training set to determine independent predictors of AKI. The study developed a nomogram based on logistic regression results to forecast AKI in UGIB. Its performance was evaluated using ROC curves, calibration curves, and decision curve analysis (DCA) in training, validation, and external validation sets.

This study incorporated 2,020 patients suffering from UGIB and 421 participants admitted to HMUSH. The Boruta algorithm identified 9 key predictors of AKI: age, creatinine-to-albumin ratio (CAR), prothrombin time (PT), alanine transaminase (AST) to aspartate aminotransferase (ALT) ratio (DeRitis), blood urea nitrogen (BUN), total bilirubin (TBIL), platelet count (PLT), pneumonia, and ascites. A nomogram was plotted utilizing these independent predictors. The nomogram demonstrated robust accuracy with area under the curve values of 0.776, 0.744, and 0.785 for the training set, validation set, and external validation set, respectively. Calibration curves disclosed good agreement between predicted and observed incidence rates of AKI. Additionally, the DCA disclosed net clinical benefit of the nomogram.

This study presented a tool developed to forecast the likelihood of developing AKI in individuals with UGIB. The tool is potentially useful for guiding timely intervention and improving clinical results.

Not applicable.