Construction and validation of a risk prediction model for secondary acute kidney injury in patients with acute upper gastrointestinal bleeding

Objective Acute upper gastrointestinal bleeding (AUGIB) is one of the most common critical conditions in clinical practice and is characterized by rapid progression and a high incidence of secondary acute kidney injury (AKI). This study aimed to analyze the clinical characteristics of patients with AUGIB, identify related risk factors for secondary AKI, and develop a predictive model for AKI risk in this patient population. Methods A retrospective analysis was conducted on 300 patients with AUGIB admitted to our department. Patients were categorized based on the occurrence of AKI within 7 days. Univariate analysis and Least Absolute Shrinkage and Selection Operator (LASSO) regression were used for feature selection, followed by multivariable logistic regression to construct a predictive model. The model’s performance was rigorously evaluated through bootstrap internal validation, calibration curves, and decision curve analysis (DCA). Results Seven independent risk factors were identified and incorporated into the SCORE-AKI: a history of renal insufficiency, hypertension, shock index > 1, Glasgow Coma Scale score ≤14, hemoglobin < 60 g/L, platelet count < 50 × 109/L, and serum creatinine > 103 μmol/L. The model showed strong discrimination with a bootstrap-corrected area under the curve (AUC) of 0.808, which was significantly superior to the Glasgow Blatchford score (AUC: 0.722, p < 0.001). The model also demonstrated excellent calibration and a positive net benefit across clinical decision thresholds. Conclusion The SCORE-AKI model is a accurate, well-calibrated, and clinically useful tool that outperforms the GBS for predicting secondary AKI risk in patients with AUGIB, potentially aiding in early risk stratification and preventive intervention.