健康管理队列慢性阻塞性肺疾病风险预测模型

发布时间：2018-06-06 13:18

本文选题：慢性阻塞性肺疾病 + 队列研究　；参考：《山东大学学报(医学版)》2017年12期

【摘要】：目的构建健康管理队列慢性阻塞性肺疾病(COPD)的风险预测模型,对模型预测效果进行评价,为COPD的防治提供理论依据。方法模型队列的建立基于"山东多中心健康管理纵向观察大数据队列",共有33 383人进入队列。通过对慢性阻塞性肺疾病发生的危险因素进行变量筛选,应用Cox比例风险回归建立COPD的预测模型,十折交叉验证法判别模型准确性,受试者工作特征曲线下面积(AUC)衡量模型的辨别能力。结果队列观察期间,共发生慢性阻塞性肺疾病136例,发病密度为118.08/10万人年。最终纳入模型的变量为年龄、性别、吸烟、白蛋白、血压和血白细胞计数。模型ROC曲线下面积为0.872(95%CI:0.810~0.934),十折交叉验证后ROC曲线下面积为0.866(95%CI:0.809~0.923)。结论构建的慢性阻塞性肺疾病预测模型在健康管理队列中有较好的预测能力。
[Abstract]:Objective to establish a risk prediction model of chronic obstructive pulmonary disease (COPD) in the health management cohort and evaluate the predictive effect of the model to provide a theoretical basis for the prevention and treatment of COPD. Methods the model cohort was established on the basis of "Shandong Multi-Center Health Management Longitudinal observation big data Cohort". A total of 33,383 people were enrolled in the queue. By screening the risk factors of chronic obstructive pulmonary disease (COPD), the predictive model of COPD was established by using Cox proportional risk regression, and the accuracy of the model was judged by 10% cross-validation. The area under the operating characteristic curve measured the discriminant ability of the model. Results during the cohort observation, 136 cases of chronic obstructive pulmonary disease (COPD) were observed, with an incidence density of 118.08 / 100 thousand person-years. The variables included in the model were age, sex, smoking, albumin, blood pressure and white blood cell count. The area under the model ROC curve is 0.87295 and the area under the ROC curve is 0.86695 CI: 0.809 / 0.9233.The area under the ROC curve is 0.87295 and 0.809 respectively, and the area under the ROC curve is 0.86695% 0.809. Conclusion the predictive model of chronic obstructive pulmonary disease has good predictive ability in health management cohort.
【作者单位】：山东大学公共卫生学院生物统计学系;山东大学齐鲁生物医学大数据中心;山东大学附属山东省肿瘤医院科教部;山东省医学科学院;
【基金】：国家自然科学基金(81273177)
【分类号】：R563.9

【相似文献】