基于生活方式相关指标的非肥胖人群非酒精性脂肪肝预测模型的构建

Construction of prediction model of non-alcoholic fatty liver disease in non-obese population based on lifestyle-related indicators

目的构建基于生活方式相关指标的非肥胖人群非酒精性脂肪肝预测模型。方法前瞻性队列研究,便利选取2021年4—10月于上海市某三级甲等综合性医院参加健康体检的992例非肥胖者,随访至2023年10月,根据随访腹部超声结果,判断受试者是否发生非酒精性脂肪肝,分为未发生组(n=834)与发生组(n=158),比较2组一般资料、体格检查、实验室检查和生活方式相关指标情况。采用LASSO回归和Logistic回归分析影响因素,构建预测模型,绘制列线图。结果通过多因素分析,最终纳入男性(OR=2.375)、年龄(OR=1.044)、高收入水平(OR=0.512)、体质量指数24.0~27.9(OR=3.556)、蛋类(OR=1.008)、蔬菜类(OR=0.998)与含糖饮料(OR=1.002)摄入量、静坐时间(OR=1.199)与睡眠时间(OR=0.661)构建预测模型。模型的ROC曲线下面积为0.846,内部验证的ROC曲线下面积为0.819,Hosmer-Lemeshow检验χ^(2)=7.478,P=0.486,Calibration校准曲线与决策曲线分析显示模型具有较好的校准度与临床实用性。结论本研究构建的预测模型具有科学性与实用性,预测效能良好,可简便、有效地预测非肥胖人群非酒精性脂肪肝的发生风险。

Objective To construct a prediction model of non-alcoholic fatty liver disease in non-obese population based on lifestyle-related indicators.Methods A prospective cohort study was conducted to select 992 non-obese subjects who participated in physical examination in a tertiary grade-A hospital in Shanghai from April to October 2021 and were followed up until October 2023.The subjects were confirmed whether they had developed non-alcoholic fatty liver disease through abdominal ultrasound during the follow-up and then they were divided into non-occurrence group(n=834)and occurrence group(n=158).General information,results of physical examination and laboratory examination and lifestyle-related indicators were compared between the two groups.LASSO regression and logistic regression were used to analyze the influencing factors and build a prediction model,and a nomogram was drawn.Results Through multi-factor analysis,male(OR=2.375),age(OR=1.044),high income level(OR=0.512),body mass index of 24.0~27.9(OR=3.556),intake of eggs(OR=1.008),vegetable(OR=0.998)and sugar-sweetened beverage(OR=1.002),sitting time(OR=1.199)and sleep time(OR=0.661)were used to construct the prediction model.The area under ROC curve of the model was 0.846,and the area under ROC curve of internal validation was 0.819.The Hosmer-Lemeshow test results ofχ^(2)=7.478,P=0.486 indicated good fit.Calibration curve and decision curve analysis showed that the model had good calibration degree and was clinically applicable.Conclusion The prediction model constructed in this study is scientific and practical,with good prediction efficiency,and can easily and effectively predict the risk of non-alcoholic fatty liver disease in non-obese population.