CHAID模型在巩义市新生儿低出生体重影响因素筛选中的应用  

THE APPLICATION OF DECISION TREE IN STUDYING THE RISK FACTORS OF LOW BIRTH WEIGHT

作　　者：张玥娇[1] 徐昕[1] 代礼[2] 邓长飞[2] 张强[1] 彭丹[1] 

机构地区：[1]四川大学华西公共卫生学院卫生统计教研室,成都610041 [2]四川大学华西二院中国出生缺陷监测中心

出　　处：《现代预防医学》2012年第1期18-20,22,共4页Modern Preventive Medicine

基　　金：国家"十一五"科技支撑计划项目(2006BAI05A01)

摘　　要：[目的]使用决策树中分类方法的CHAID算法模型和非条件Logistic回归模型,探讨新生儿低出生体重危险因素,并比较决策树的和Logistic模型间结果的异同。[方法]根据新生儿是否低出生体重分类。分别建立决策树CHAID模型和非条件Logistic模型。[结果]母亲孕周是新生儿低出生体重最主要的影响因素,在孕周小于37周的亚群中,多胎的新生儿发生低出生体重的风险更高。[结论在二分类结果变量的危险因素分析中,非条件Logistic模型和决策树模型都有较高的应用价值,两种方法可以相互结合,得到更为全面的结果。[Objective]To apply CHAID model of decision tree,and Logistic regression model to discuss the risk factor of low birth weight.In addition,to compare the differences and similarities between the two models and explore the useful information for Logistic regression model from decision tree.[Methods]With the dependent variable of birth weight,established Logistic regression model and decision tree model.[Results]The potential risk factors of low birth weight in infants included weeks of pregnancy and multiple pregnancies.[Conclusion]Both Logistic regression and decision tree are useful in studying risk factors of binary dependent variable.And when these two different methods are combined,it is much more helpful to make the good use of the information to find out the risk factors.

关 键 词：影响因素 低出生体重 CHAID 

分 类 号：R195.1[医药卫生—卫生统计学]