超重及肥胖人群高尿酸血症发病风险预测模型的构建及评价  

Construction and evaluation of hyperuricemia prediction model for overweight and obese population

作　　者：李禄伟 黄倩 施佳成 刘晓玲[1] 王彩梅[2] 于萍[1] 吴岚[3] 覃洋 江仁美[1] 于健[1] LI Luwei;HUANG Qian;SHI Jiacheng;LIU Xiaoling;WANG Caimei;YU Ping;WU Lan;QIN Yang;JIANG Renmei;YU Jian(Department of Endocrinology,Guilin Medical College Hospital,Guilin 541001,China;不详)

机构地区：[1]桂林医学院附属医院内分泌科,广西桂林541001 [2]桂林医学院附属医院检验科,广西桂林541001 [3]桂林医学院附属医院神经内科,广西桂林541001

出　　处：《实用医学杂志》2021年第18期2401-2406,共6页The Journal of Practical Medicine

基　　金：广西医疗卫生适宜技术开发与推广应用项目(编号:S2019062);桂林市科学研究与技术开发计划项目(编号:20190218-5-1)。

摘　　要：目的利用分类树和列线图构建超重及肥胖人群高尿酸血症发病风险预测模型并进行评价。方法选取超重及肥胖者5098例为研究对象,随机抽取3582例(70%)超重及肥胖者构成建模组,剩余1516例(30%)构成验证组进行内部验证。两组均以高尿酸血症有无分为病例组和对照组(血清尿酸≥420μmol/L定义为高尿酸血症),对病例组和对照组人群临床代谢特征进行比较并构建分类树模型和列线图模型,最后使用ROC曲线、DCA曲线、CIC曲线对两个预测模型进行比较,分析其临床实用性。结果训练集和验证集中分类树及列线图模型均提示男性、Cr、TG、年龄、HDL-c、LDL-c等6个变量是超重及肥胖人群高尿酸血症的影响因素。两组两个模型ROC曲线均提示有中度预测价值,DCA曲线患病风险概率在约为0.1~0.7范围内,两个模型的净受益率都高于0,而CIC曲线提示两个模型均有一定的临床影响价值。结论分类树和列线图两个预测模型筛选的影响因素包括男性、Cr、TG、年龄、HDL-c、LDL-c,两个模型均具有一定的预测价值和临床实用性。Objective To construct and evaluate the hyperuricemia prediction model for overweight and obese people.Methods A retrospective study was conducted to collect 5098 overweight and obese people as research subjects.Among of them,3582(70%)were randomly selected to form the modeling group,and the rest 1516(30%)as the internal validationgroup.The subjects in the modeling group and internal validation group were divided into case group and control group according to the presence or absence of hyperuricemia(serum uric acid≥420μmol/L).The case group and control group were compared in terms of clinical metabolic features and classification tree and nomogram models were constructed.Finally,ROC curve,DCA curve and CIC curve were used to compare the two prediction models and analyze their clinical practicability.Results The classification tree and nomogram models of training set and validation set showed that male,Cr,TG,age,HDL-c,LDL-c were the influencing factors of hyperuricemia among the overweight and obese people.The ROC curve of the two models in the two groups indicated moderate predictive value.The risk probability of DCA curve was in the range of 0.1~0.7,and the net benefit rate of the two models was higher than 0.The CIC curve indicated that the two models had certain clinical impact value.Conclusion The influencing factors predicted from the classification tree and nomogram models are male,Cr,TG,age,HDL-c and LDL-c,and both models are ofsome value for prediction and clinical practicability.

关 键 词：超重及肥胖 高尿酸血症 分类树模型 列线图模型 预测模型 ROC曲线 DCA曲线 CIC曲线 

分 类 号：R589.9[医药卫生—内分泌]