An analysis on the influencing factors of pregnant women's choice of abortion methods by the combination of decision tree and logistic regression model
- VernacularTitle:决策树结合 Logistic 回归分析妊娠妇女选择人工流产方式的影响因素
- Author:
Fu-Dong LI
1
;
Li-Li HUANG
;
Yan-Jin YU
;
Fan HE
;
Yi SHEN
Author Information
1. 浙江省疾病预防控制中心
- Keywords:
Induced abortion;
Influencing factors;
Decision tree;
Logistic regression
- From:
Journal of Preventive Medicine
2015;(4):328-333
- CountryChina
- Language:Chinese
-
Abstract:
Objective To understand the influencing factors of pregnant women's choice of abortion methods by exhaustive CHAID decision tree and logistic regression model,and to evaluate the two statistical methods.Methods Pregnant women who underwent artificial abortion in 2011 were selected from 30 medical institutions of 10 cities in Zhejiang province.And a questionnaire survey was conducted.The influencing factors of pregnant women's choice of abortion methods were analyzed by exhaustive CHAID decision tree and logistic regression model,and the two methods were compared.Results Exhaustive CHAID tree analysis showed that the choice of artificial abortion methods were associated with preferences,gestational age, induced abortion history,marital status,fertility condition and abortion cognitive such as slight pain in drug abortion,the time limitation of drug abortion,greater possibility of curettage after induced abortion.Preferences and gestational age were the two major influencing factors.The model was stable and well fitting with the risk estimation 0.119 and prediction error 11.90%.And the logistic regression analysis showed the similar results expect induced abortion history and marital status. Age,education level and decision source were also the influencing factors in the logistic regression model.Conclusion The combination of decision tree and logistic regression analysis can analysis the influencing factors of pregnant women's choice of abortion methods,and information from the two methods should be combined to improve the efficiency of analysis.
