Objective To investigate the clinical application value of predicting the severity of acute pancreatitis(AP)based on MRI-T2WI radiomics nomogram.Methods A total of 375 patients with AP were analyzed retrospectively,who were divided into 281 cases in the training group and 94 cases in the validation group according to the ratio of 3∶1.Based on MRI-T2WI image,man-ual segmentation was performed for the pancreatic parenchyma.The radiomics feature were selected by feature extraction and dimen-sionality reduction,the support vector machine(SVM)classifier were used to construct the radiomics model.Logistic regression analysis was used to screen out independent risk factors,and an radiomics nomogram model was constructed in combined with the Radiomics score(Radscore),and the predictive performances of the models were evaluated.Results Receiver operating characteristic(ROC)curve analysis showed that the predictive efficacy of radiomics nomogram model[training group,area under the curve(AUC)=0.893;val-idation group,AUC=0.889]was higher than that of clinical model(training group,AUC=0.799;validation group,AUC=0.809)and radiomics model(training group,AUC=0.814;validation group,AUC=0.823).Conclusion The radiomics nomogram based on MRI-T2WI radiomics features and independent risk factors has high clinical application value for the prediction of AP severity.