基于机器学习的非酒精性脂肪性肝病合并高脂血症性急性胰腺炎严重程度预测模型构建与验证OA

Objective To construct a machine learning-based predictive model for the severity of non-alcoholic fatty liver disease(NAFLD)complicated with hyperlipidemic acute pancreatitis(HLAP),and validate and evaluate the predictive performance of the model.Methods Clinical data of 396 patients with NAFLD complicated with HLAP admitted to Zhengzhou Central Hospital Affiliated to Zhengzhou University from January 2018 to March 2025 were retrospectively collected.The data were randomly divided into a training set and a test set at a ratio of 7∶3.In the training set,the least absolute shrinkage and selection operator(LASSO)and multivariate logistic regression were used to screen characteristic predictors,and the optimal algorithm was selected through multi-model comparison to reconstruct the predictive model.Model perfor-mance was evaluated using the area under the ROC curve(AUC),with model validity further assessed by calibration curves and decision curve analysis(DCA),followed by importance ranking and visual interpretation of the feature predictors.Results A total of five characteristic predictors were screened and ranked in descending order of importance:ascites,serum calcium ion,lactate dehydrogenase,alanine aminotrans-ferase,and hemoglobin.The Gaussian Naive Bayes(GNB)prediction model con-structed based on these factors showed superior performance,with AUC values of 0.909 in the training set,0.901 in the validation set,and good performance in the test set(AUC=0.883).Conclusion The machine learning-based GNB prediction model can identify disease severity in patients with NAFLD complicated by HLAP at an early stage,demonstrating good predictive performance and clinical applicability.