基于CT和临床特征的列线图模型预测急性胰腺炎经口进食时机:多中心研究OA

Objective:To identify factors affecting oral feeding resumption timing in acute pancreatitis(AP)and establish a nomogram model that integrates clinical indicators and CT gastrointestinal tract features for guiding AP feeding decisions.Methods:A retrospective multicenter study enrolled 152 AP patients(randomized to 60%training and 40%internal validation cohorts)plus two external validation patients(n=59 and n=55,respectively)from the other two tertiary hospitals.Patients were stratified by early or late feeding initiation(within or after 7 days onset).Clinical data,CT findings including Extrapancreatic Inflammation on CT(EPIC)scores,and gastrointestinal change on CT scores were collected.Independent predictors of early feeding were identified via univariate analysis and multivariate Logistic regression to build a nomogram,with performance evaluated by using the receiver operating characteristic curve,calibration curve,and decision curve analysis(DCA).Results:Serum albumin,EPIC score,and gastrointestinal change on CT score were independent predictors of early oral feeding(P<0.05).The AUC of nomogram model was 0.80(training cohort),outperforming clinical-only(AUC=0.70)and imaging-only(AUC=0.78)models,internal and the other two external validation AUCs were 0.68,0.88 and 0.84,respectively.Calibration and DCA verified good consistency and clinical utility.Conclusion:The integrated nomogram model accurately predicts AP oral feeding time,providing valuable guidance for personalized nutritional management.