Abstract: Objective To develop a predictive model for acute kidney injury (AKI) in liver cirrhosis patients within 7 days of hospitalization, and to identify independent risk factors for AKI, and offer evidence for early clinical identification of high-risk patients and intervention strategy development. Methods A total of 157 liver cirrhosis patients admitted from January 2023 to October 2025 were retrospectively enrolled in the 904 Hospital of the Joint Logistics Support Force of the People′s Liberation Army. They were divided into an AKI group (n=69) and a non-AKI group (n=88) based on AKI occurrence within 7 days of hospitalization. Clinical features and laboratory parameters were gathered and analyzed comparatively across groups. The least absolute shrinkage and selection operator (LASSO) regression was employed to identify potential risk factors, and the screened variables were in turn analyzed through multivariate logistic regression to ascertain independent risk factors. A predictive model was further constructed, and the effectiveness of this model was evaluated by means of the receiver operating characteristic (ROC) curve and corresponding area under the curve (AUC). Results Univariate analyses revealed marked intergroup disparities in age, Child-Pugh classification, incidence of ascites and spontaneous bacterial peritonitis, rate of diuretic administration, as well as multiple laboratory parameters (white blood cell count, serum creatinine, estimated glomerular filtration rate, procalcitonin), with all variations achieving statistical significance (P<0.05). LASSO regression identified four key variables including age, hemoglobin, estimated glomerular filtration rate, and serum creatinine. Multivariate logistic regression confirmed age (OR=1.180, 95%CI:1.034-1.347) and serum creatinine (OR=1.072, 95%CI:1.015-1.132) as independent AKI risk factors (P<0.05), while hemoglobin (OR=0.909, 95%CI:0.841~0.982) and estimated glomerular filtration rate (OR=0.921, 95%CI:0.871~0.974) were independent protective factors (P<0.05). The combined model achieved an AUC of 0.972 (95%CI:0.940~1.000) with 97.1% sensitivity and 95.5% specificity, outperforming individual indicators. Conclusion The LASSO-derived combined model effectively forecasts 7-day AKI risk among cirrhotic patients. Age, serum creatinine, hemoglobin and estimated glomerular filtration rate act as pivotal predictive indicators, facilitating timely clinical vigilance and tailored intervention to optimize patient prognosis.

Key words: Liver cirrhosis, Acute kidney injury, LASSO regression, Predictive model, Risk factor