健康管理人群慢性肾脏病风险预测模型

发布时间：2019-04-11 11:06

【摘要】：目的构建健康管理人群慢性肾脏病(CKD)发病风险预测模型。方法从山东多中心健康管理纵向观察队列中选取年龄20岁以上、至少有两次纵向观察结果、初次观察结果未患CKD的健康管理对象,共17 654人,随访观察结局为CKD。应用Cox比例风险回归模型建立风险预测模型,以受试者工作特征曲线下面积(AUC)评价模型的拟合效果,应用十折交叉验证法验证模型的稳定性。结果观察期间共有770例新发CKD病例,发病密度为17.69/1 000人年。最终纳入模型的预测因子有年龄、性别、高血压、糖尿病、血肌酐、血尿素氮、血尿酸、嗜碱性粒细胞百分比。模型AUC为0.685(95%CI:0.678~0.692),且稳定性较好。结论建立的CKD风险预测模型在健康管理人群中有较好的预测能力。
[Abstract]:Objective to establish a risk prediction model for chronic kidney disease (CKD) in health management population. Methods from the longitudinal observation cohort of multi-center health management in Shandong Province, 17 654 health management subjects with no CKD were selected from the longitudinal observation cohort aged over 20 years with at least two longitudinal observations. The outcome of follow-up observation was CKD.. The Cox proportional hazard regression model is used to establish the risk prediction model. The stability of the model is verified by the fitting effect of the area (AUC) model under the receiver operating characteristic curve and the application of 10% cross-validation to verify the stability of the model. Results there were 770 new cases of CKD during the observation period, and the incidence density was 17.69 / 1 000 person-years. The predictors included in the model were age, sex, hypertension, diabetes, serum creatinine, blood urea nitrogen, serum uric acid, and percentage of basophils. The model AUC is 0.685 (95%CI:0.678~0.692) and the stability is good. Conclusion the CKD risk prediction model has good predictive ability in health management population.
【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研究中心;北大医疗淄博医院;北大医疗淄博医院社会工作部;北大医疗淄博医院体检中心;
【基金】：国家自然科学基金(81273177)
【分类号】：R692

【相似文献】