主成分Logistic回归在筛选冠心病危险因素中的应用  

Application of Logistic Regression Based on Principal Component Analysis in the Screening of Risk Factors of Coronary Heart Disease

作　　者：行岳真[1] 祁瑞[1] 李金梅[1] 李贞子[1] 王丽敏[1] 孙宏[1] 刘艳[1] 马葆华[1] 隋虹[1] 

机构地区：[1]哈尔滨医科大学卫生统计学教研室,黑龙江哈尔滨150081

出　　处：《实用预防医学》2012年第8期1138-1140,共3页Practical Preventive Medicine

基　　金：哈尔滨市科技创新人才研究专项资金项目(2006RFQXS072);黑龙江省卫生厅项目(2003130)

摘　　要：目的探讨主成分Logistic回归方法在筛选冠心病危险因素中的应用。方法选择2009年4月-2010年8月在哈尔滨医科大学第一附属医院行冠脉造影术者,按照金标准分为冠心病组(465例)和对照组(277例);首先对Lo-gistic回归模型进行共线性诊断,然后应用主成分改进的Logistic回归分析,得到并解释最终的回归模型。结果共线性诊断提示各变量之间存在明显的共线性,采用主成分改进的Logistic回归分析显示,冠心病与年龄、性别、载脂蛋白A、TG、HDL-C、糖尿病史、高血压史、吸烟史在内的多种因素有关。结论主成分改进的Logistic回归在筛选冠心病危险因素中具有较好的作用,在对疾病危险因素进行Logistic回归分析时,若多变量间存在多重共线性,采用主成分改进的Lo-gistic回归分析能得出更好的回归模型。Objective To explore the application of the logistic regression based on principal component analysis in the screening of risk factors of coronary heart disease （CHD）. Methods A total of 742 patients who received coronary angiogra- phy in the First Affliated Hospital of Harbin Medical University during April 2009 and August 2010 were recruited in the study. The patients were classified into the CHD group （n= 465） and the control group （n= 277） according to the gold standard diag- nosing CHD. After making multi - collinearity diagnosis on logistic regression model, the logistic regression model improved by principal component analysis was used to get and explain a new model. Results There was significant multi - collinearity among variables. The logistic regression model improved by principal component analysis showed that CHD was correlated with many factors such as age, gender, apoA, TG, HDL- C, diabetes history, hypertension history and smoking history. Conclu- sions The logistic regression model improved by principal component analysis is applicable in the screening of risk factors of CHD. It can provide a better regression model if there is multi - collinearity among variables.

关 键 词：冠心病 危险因素 主成分 LOGISTIC回归 

分 类 号：R541.4[医药卫生—心血管疾病]