基于group lasso logistic回归的6月龄婴儿贫血预测模型的构建  

Construction of predictive model for anemia in 6-month-old infants on the basis of group lasso logistic regression

作　　者：张啸 赵薇 胡焕青[2] 黄爱群 陈大方[1] ZHANG Xiao;ZHAO Wei;HU Huan-qing;HUANG Ai-qun;CHEN Da-fang(Department of Epidemiology and Biostatistics,School of Public Health,Peking University,Beijing 100191,China)

机构地区：[1]北京大学公共卫生学院流行病与卫生统计学系,北京100191 [2]中国疾病预防控制中心妇幼保健中心

出　　处：《中国慢性病预防与控制》2020年第3期199-205,共7页Chinese Journal of Prevention and Control of Chronic Diseases

摘　　要：目的构建6月龄婴儿中度贫血的预测模型,为婴儿贫血的预防和控制提供更好的指导。方法于2015年9月至2019年6月,基于方便抽样,选择在河北省石家庄市正定县、辽宁省鞍山市立山区、湖南省岳阳市岳阳县以及厦门市海沧区和集美区生活的2431名孕晚期孕妇进行问卷调查,并对其分娩的婴儿进行随访观察。采用SAS 9.4软件进行数据描述和χ^2检验。使用R 3.6.1软件拟合全变量进行group lasso logistic回归,筛选6月龄婴儿中度贫血的影响因素及构建预测模型,并绘制受试者工作特征(ROC)曲线,计算灵敏度、特异度和曲线下面积(AUC)评价模型的预测效果。结果2431名6月龄婴儿中,总贫血率为24.60%,轻度贫血率为22.46%,中度贫血率为2.14%。以中度贫血为结局,最终筛选出11项有意义的指标,分别为低出生体重、分娩方式、维生素D摄入情况、当前母乳喂养、最近两周是否感冒、婴儿贫血史、母亲教育程度、孕中晚期是否吃牛(羊)肉、孕中晚期是否吃鸡(鸭)等禽类肉、孕中晚期是否吃海带(紫菜)等海产品和地区。模型的AUC值为0.7832(95%CI:0.7228~0.8435),若取阈值为1.77%,则灵敏度为90.38%,特异度为45.65%。结论采用group lasso logistic回归构建的预测模型具有较好的预测能力,当阈值为1.77%时,能较好地发现6月龄婴儿中度贫血的高危人群。Objective To construct a predictive model of moderate anemia in 6-month-old infants,and to provide the better guidance for prevention and control of infant anemia.Methods From September of 2015 to June of 2019,2431 third trimester pregnant women from 5 districts(counties)of China were investigated by questionnaire,and the infants were followed up.Data description andχ^2 test were performed using SAS 9.4 software.Group lasso logistic regression was performed using R 3.6.1 software to screen influencing factors of moderate anemia in 6-month-old infants and to construct the predictive model,and the receiver operating characteristic(ROC)curve was drawn,the sensitivity,specificity and the area under the curve(AUC)were calculated to evaluate the predictive model.Results In 24316-month-age infants,the total anemia rate was 24.60%,light anemia rate was 22.46%,moderate anemia rate was 2.14%.On the basis of moderate anemia,11 meaningful indicators were selected to construct predictive model,including low birth weight,mode of delivery,vitamin D intake,current breastfeeding,whether to have a cold within two weeks,history of anemia in infants,education level of mother,whether to eat beef(mutton)in middle or late pregnancy,whether to eat poultry in middle or late pregnancy,whether to eat kelp(seaweed)and other seafood in middle or late pregnancy,region.AUC of the model was 0.7832(95%CI:0.7228-0.8435),if 1.77%served as threshold,the sensitivity was 90.38%and specificity was 45.65%.Conclusion The predictive model constructed by group lasso logistic regression has better predictive effects and when the threshold is 1.77%,the high-risk groups of infants with moderate anemia can be found.

关 键 词：婴儿 贫血 预测模型 

分 类 号：R725.5[医药卫生—儿科]