非肥胖型代谢功能障碍相关脂肪性肝病诊断模型的建立与验证  

Construction and validation of the diagnostic model for metabolic dysfunctionassociated steatotic liver disease among the non-obese population

作　　者：王玉丰 胡旻萱 李云涛[1] 朱可欣 季国忠[1] 袁源[2] WANG Yufeng;HU Minxuan;LI Yuntao;ZHU Kexin;JI Guozhong;YUAN Yuan(Department of General Practice,The Second Affiliated Hospital of Nanjing Medical University,Nanjing 210000,China;Health Management Centre,The Second Affiliated Hospital of Nanjing Medical University,Nanjing 210000,China)

机构地区：[1]南京医科大学第二附属医院全科医学科,南京210000 [2]南京医科大学第二附属医院健康管理中心,南京210000

出　　处：《医学新知》2025年第2期141-150,共10页New Medicine

基　　金：江苏省卫健委面上项目(M2022045)。

摘　　要：目的分析非肥胖型代谢功能障碍相关脂肪性肝病(metabolic dysfunctionassociated steatotic liver disease,MASLD)的预测指标并构建诊断模型。方法回顾性分析2022年8月至2024年7月南京医科大学第二附属医院健康管理中心体检人群的相关资料,根据体检时间将2022年8月至2024年5月完成体检的人群设为建模组,2024年6月至2024年7月完成体检的人群设为验证组,采用Lasso回归筛选潜在预测指标,二元Logistic回归分析确定指标并构建列线图,采用混淆矩阵、受试者工作特征(ROC)曲线及其曲线下面积(AUC)、校准曲线分析(CCA)、决策曲线分析(DCA)评价模型效能。结果共纳入791例体检对象,其中建模组607例,验证组184例;非肥胖型MASLD 292例,患病率为36.92%。多因素Logistic回归分析显示,体重指数[OR=1.860,95%CI(1.559,2.219)]、空腹血糖[OR=1.415,95%CI(1.174,1.707)]、甘油三酯[OR=1.308,95%CI(1.021,1.675)]、γ-谷氨酰转移酶[OR=1.012,95%CI(1.005,1.020)]、尿酸与高密度脂蛋白胆固醇比值[OR=1.004,95%CI(1.002,1.007)]为非肥胖型MASLD的预测指标,建模组和验证组的准确率分别为74.0%和72.8%,精确率分别为67.7%和72.7%,AUC分别为0.814[95%CI(0.780,0.848)]和0.819[95%CI(0.755,0.883)],建模组和验证组Hosmer-Lemeshow检验均无统计学意义(P>0.05),拟合优度较好,CCA分析显示模型“预测概率”和“实际概率”的一致性较好,DCA显示模型具有较好的净收益。结论本研究构建的诊断模型对非肥胖型MASLD具有良好诊断能力,可用于BMI正常人群的早期筛查。Objective To analyze the predictive indicators of non-obese metabolic dysfunctionassociated steatotic liver disease(MASLD)and construct a diagnostic model.Methods A retrospective analysis was conducted on data from individuals who underwent health examinations at the Health Management Center of the Second Affiliated Hospital of Nanjing Medical University between August 2022 and July 2024.The study population was divided into a modeling group(those who completed examinations between August 2022 and May 2024)and a validation group(those who completed examinations between June 2024 and July 2024).Lasso regression was used to screen potential predictive indicators,and binary Logistic regression was employed to identify key indicators and construct a nomogram.Model performance was evaluated using a confusion matrix,receiver operating characteristic(ROC)curve analysis with area under the curve(AUC),calibration curve analysis(CCA),and decision curve analysis(DCA).Results A total of 791 physical examination subjects were included,with 607 cases in the modeling group and 184 cases in the validation group.Among them,292 cases were non-obese MASLD,with a prevalence of 36.92%.Multivariate Logistic regression analysis identified body mass index(BMI)[OR=1.860,95%CI(1.559,2.219)],fasting blood glucose[OR=1.415,95%CI(1.174,1.707)],triglyceride[OR=1.308,95%CI(1.021,1.675)],gamma-glutamyl transferase[OR=1.012,95%CI(1.005,1.020)],and the uric acid to high-density lipoprotein cholesterol ratio[OR=1.004,95%CI(1.002,1.007)]as predictive indicators for non-obese MASLD.The accuracy rates for the modeling and validation groups were 74.0%and 72.8%,respectively,while the precision rates were 67.7%and 72.7%,respectively.The AUC values were 0.814[95%CI(0.780,0.848)]for the modeling group and 0.819[95%CI(0.755,0.883)]for the validation group.The Hosmer-Lemeshow test showed no statistical significance(P>0.05)for both groups,indicating good model fit.CCA demonstrated strong agreement between predicted and actual probabilities,and DCA indicated

关 键 词：代谢功能障碍相关脂肪性肝病 非肥胖 诊断模型 LOGISTIC回归 预测模型 

分 类 号：R575.5[医药卫生—消化系统]