基于LASSO-Logistic回归的脓毒症相关急性肾损伤患者死亡风险预测OA

Objective To investigate the factors influencing 30-day mortality in patients with sepsis-associated acute kidney injury(SAAKI)and to develop a predictive model based on LASSO-Logistic regression,providing a reference for early identification of high-risk patients in clinical practice.Methods A retrospective study was conducted on 240 SAAKI patients admitted to the Intensive Care Unit at the Fifth Hospital of Xiamen between January 2021 and August 2025.Based on their 30-day survival outcomes,patients were divided into a non-survival group(n=60)and a survival group(n=180).Baseline clinical characteristics were compared between the two groups.Multivariate logistic regression analysis was performed to identify key factors associated with 30-day mortality in SAAKI patients.Based on these findings,a nomogram was constructed for individualized risk prediction.Subsequently,the model's discriminative ability,calibration accuracy,and clinical utility were systematically evaluated using receiver operating characteristic(ROC)curve analysis,calibration curve analysis,and decision curve analysis(DCA).Results Compared with the survival group,the non-survival group had significantly higher values in age,proportions of smoking,diabetes,and hypertension,as well as higher heart rate,respiratory rate,blood lactate,proportion of stage Ⅲa acute kidney injury,multiple organ dysfunction syndrome(MODS),mechanical ventilation,vasoactive agent use,and acute physiology and chronic health evaluationⅡ(APACHEⅡ)score.Conversely,mean arterial pressure,white blood cell count,neutrophil count,fibrinogen level,and oxygenation index were significantly lower in the death group(P<0.05).Multivariate logistic regression analysis identified age,APACHEⅡ score,MODS,and mechanical ventilation as independent factors for 30-day mortality in SAAKI patients,whereas neutrophil count and oxygenation index were independent protective factors against death within one month after admission(OR=1.151,95%CI:1.099-1.205,P<0.001;OR=1.748,95%CI:1.488-2.054,P<0.001;OR=9.333,95%CI:4.321-20.159,P<0.001;OR=4.896,95%CI:1.126-21.286,P=0.034;OR=0.499,95%CI:0.409-0.608,P<0.001;OR=0.908,95%CI:0.879-0.938,P<0.001).ROC curve analysis indicated that the area under the curve(AUC)was 0.835(95%CI:0.771-0.897)in the training set and 0.775(95%CI:0.635-0.893)in the validation set.DCA revealed that the nomogram prediction model had favorable clinical utility.Calibration curve analysis demonstrated a strong concordance between the model's predicted probabilities and the actual observed event rates.Single-sample prediction results suggested that oxygenation index,APACHEⅡ score,and neutrophil count contributed the most to the model,with the predicted probability of death reaching as high as 96.4%.Conclusions Age,neutrophil count,oxygenation index,APACHEⅡ score,MODS,and mechanical ventilation are independent factors influencing 30-day mortality in SAAKI patients.The predictive model constructed based on LASSOlogistic regression demonstrates good discriminative ability and clinical utility.Combined with individual risk prediction,it enables personalized risk assessment and provides an important reference for the early identification of highrisk patients.