基于体检队列的2型糖尿病风险预测模型

发布时间：2018-01-11 09:16

  本文关键词：基于体检队列的2型糖尿病风险预测模型　出处：《山东大学学报(医学版)》2016年09期 　论文类型：期刊论文

  更多相关文章： 型糖尿病 队列 风险预测模型 体检

【摘要】：目的构建体检者2型糖尿病发病风险预测模型。方法选择2005年1月至2010年12月在山东大学附属省立医院、山东大学附属千佛山医院体检中心体检的非糖尿病者16 715人,随机选取70%体检者为训练组,用于建立Cox预测模型,逐步选择法进行变量选择,使用十折交叉验证法检验模型的稳定性,根据预后指数制定风险分级;剩余30%的体检者为校验组,对模型进行组外验证,再次评价模型效果。结果观察期间共新发生2型糖尿病858例,发病密度为15.14‰。最终纳入模型的变量包括年龄、体质量指数、空腹血糖、甘油三酯、是否患高血压以及白细胞自然对数;训练组ROC曲线下面积为0.742(95%CI:0.732~0.752),校验组ROC曲线下面积为0.760(95%CI:0.748~0.772)。结论建立的2型糖尿病风险预测模型在体检者中有较好的预测能力。
[Abstract]:Objective to establish a model for predicting the risk of type 2 diabetes mellitus (T2DM) in physical examiners. Methods from January 2005 to December 2010, we selected the provincial hospital affiliated to Shandong University. A total of 16 715 non-diabetic patients in the physical examination center of Qianfushan Hospital affiliated to Shandong University were selected randomly as training group, which was used to establish Cox prediction model and to select variables by stepwise selection method. The stability of the model was tested by a ten fold cross validation method and the risk classification was determined according to the prognostic index. The remaining 30% physical examination group was the check group, the model was verified outside the group, and the model effect was evaluated again. Results during the observation period, 858 cases of new type 2 diabetes mellitus occurred. The incidence density was 15.14 鈥，

本文编号：1408949