基于套索回归与逻辑回归构建肥胖儿童青少年胰岛素抵抗的预测模型

Biological prediction model of insulin resistance in obese adolescents based on LASSO-Logistic regression

目的 了解影响肥胖儿童青少年胰岛素抵抗发生的风险因子,并构建预测胰岛素抵抗发病风险的列线图模型。方法 于2019—2021年通过某减肥训练营随机招募404名10~17岁肥胖儿童青少年,进行血脂和胰岛素抵抗的检测和诊断,通过套索回归筛选影响胰岛素抵抗发生的关键血液指标,通过逻辑回归构建列线图模型,通过Bootstrap法进行模型内部验证,通过受试者工作特征曲线下面积(ROC-AUC)和临床决策曲线对列线图模型进行校准度和稳定性评估。结果 针对肥胖儿童青少年所构建的胰岛素抵抗发病风险预测模型包括甘油三酯、低密度脂蛋白胆固醇、总胆固醇/高密度脂蛋白胆固醇3个风险因子。AUC=0.825(95%CI 0.782~0.868);内部验证结果C-Index为0.804;列线图模型预测胰岛素抵抗发生风险的平均绝对误差为0.015,Brier得分为0.163。Hosmer-Lemeshow拟合优度检验显示预测模型效果良好(χ^(2)=5.59,P=0.70)。结论 基于LASSO-Logistic回归构建的基于甘油三酯、低密度脂蛋白胆固醇、总胆固醇/高密度脂蛋白胆固醇的多指标联合的列线图模型,能有效预测肥胖儿童青少年胰岛素抵抗的发病风险。

OBJECTIVE To identify risk factors affecting the development of insulin resistance in obese adolescents,and to build a nomograph model for predicting the risk of insulin resistance and achieve early screening of insulin resistance.METHODSA total of 404 obese adolescents aged 10 to 17 years were randomly recruited through a weight loss camp for the detection and diagnosis of lipids and insulin resistance between 2019 and 2021,and key lipid indicators affecting the development of insulin resistance were screened by Lasso regression,nomogram model was constructed,and internal validation of the models was performed by Bootstrap method,and the area under the working characteristic curve(ROC-AUC) and clinical decision curve were used to assess the calibration degree and stability of the column line graph.RESULTS The AUC was 0.825(95% CI 0.782-0.868),the internal validation result C-Index was 0.804,the mean absolute error of the column line graph model to predict the risk of insulin resistance was 0.015 and the Brier score was 0.163.The Hosmer-Lemeshow goodness-of-fit test showed that model is ideal and acceptable(χ^(2)=5.59,P=0.70).CONCLUSIONThe nomogram model of triglyceride,low-density lipoprotein cholesterol and total cholesterol/high-density lipoprotein cholesterol based on Lasso-logistic regression can effectively predict the risk of insulin resistance in obese children and adolescents.