妊娠期亚临床甲状腺功能减退症的决策树模型构建

Construction of decision tree model for subclinical hypothyroidism in pregnancy

目的:探讨妊娠期合并亚临床甲状腺功能减退症(SCH)的危险因素,并构建决策树模型。方法:回顾性分析261例产妇的临床资料,采用Logistic回归分析筛选SCH发生的危险因素,运用Modeler软件构建决策树模型。结果:SCH发生率为12.64%。Logistic回归分析结果显示,年龄≧30岁、BMI<24kg/m2、初产、妊娠期糖尿病、子痫前期、贫血均是妊娠期产妇合并SCH的影响因素(P<0.05)。决策树模型显示,贫血是妊娠期产妇合并SCH最为重要的影响因素。ROC结果显示,多因素的AUC值高于决策树(Z=2.470,P=0.140)。结论:年龄、BMI、孕产史、妊娠期糖尿病、子痫前期、贫血均是妊娠期产妇合并SCH的影响因素,基于上述因素构建的决策树风险预测模型具有较好的风险预测效能。

Objective:To investigate the risk factors of subclinical hypothyroidism(SCH)during pregnancy and establish a decision tree model.Methods:The clinical data of 261 parturient women were retrospectively analyzed,the risk factors of SCH were screened by Logistic regression analysis,and the decision tree model was constructed by Modeler software.Results:The incidence of SCH was 12.64%.Logistic regression analysis showed that age≥30 years old,BMI<24kg/m2,first delivery,gestational diabetes mellitus,preeclampsia and anemia were all influential factors for SCH in pregnant women(P<0.05).Decision tree model shows that anemia is the most important factor affecting maternal SCH during pregnancy.ROC results showed that the AUC value of multi-factor was higher than that of decision tree(Z=2.470,P=0.140).Conclusion:Age,BMI,pregnancy and childbirth history,gestational diabetes,preeclampsia,and anemia are all influencing factors for SCH during pregnancy.The decision tree risk prediction model constructed based on these factors has good risk prediction performance.