目的 构建基于脐血胆红素水平结合临床危险因素预测ABO新生儿溶血病（HDN）发生高胆红素血症风险的预测模型并进行效能验证。 方法 选择500例ABO-HDN患儿作为建模集，100例ABO-HDN患儿作为验证集，根据是否发生新生儿高胆红素血症，分为高胆红素血症组（NHB）和非高胆红素血症组（Non-HB），检测其脐血胆红素水平，并收集患儿和产妇的一般资料及临床资料。运用多因素logistic回归分析以及ROC曲线，评价脐血胆红素水平结合临床危险因素预测模型对ABO-HDN发生NHB风险的预测效能。 结果 建模集500例ABO-HND患者中，发生NHB者有225例，发生率为45%。单因素分析显示的影响因素有：母亲年龄，胎膜早破，妊娠期合并高血压、糖尿病、甲状腺疾病及贫血，宫内窘迫，头颅血肿，感染，胎龄，新生儿窒息，喂养方式，低出生体重，颅内出血和呼吸窘迫（P < 0.05）。多因素logistic回归分析显示：早产、新生儿窒息、感染、G-6-PD缺乏、颅内出血和呼吸窘迫为独立的危险因素（OR = 2.377，7.457，54.438，45.487，2.788，5.407，P < 0.05）。脐血胆红素水平、临床危险因素、脐血胆红素水平结合临床危险因素3种模型预测ABO-HDN发生NHB的风险，ROC曲线下面积AUC值分别为：0.817（95%CI：0.776 ~ 0.859），0.876（95%CI：0.845 ~ 0.907），0.944（95%CI：0.921 ~ 0.966），均P < 0.05。将验证集100例ABO-HDN患者资料代入脐血胆红素水平结合临床危险因素的预测模型进行效能验证，ROC曲线下面积AUC值为0.951（95%CI：0.776 ~ 0.859），P < 0.05。 结论 脐血胆红素水平结合临床危险因素预测模型能准确的预测ABO-HDN发生NHB的风险，可为ABO-HDN发生NHB的预防、诊疗和规范化管理提供参考依据。

Objective To construct and verify a predictive model based on bilirubin levels in umbilical cord blood and clinical risk factors to predict the risk of hyperbilirubinemia in ABO neonatal hemolysis disease. Methods A total of 500 ABO-HDN children were selected as the modeling set. 100 ABO-HDN children served as the verification set. Depending on whether neonatal hyperbilirubin was occurring， they were divided into hyperbilirubinemia group and non-hyperbilirubin group（Non-HB）. The bilirubin level in umbilical cord blood was detected and the general and clinical data of children and mothers were collected. Multivariate Logistic regression analysis and ROC curve were used to evaluate the prediction efficiency of the prediction model of bilirubin levels in umbilical cord blood combined with clinical risk factors for the risk of NHB in ABO-HDN. Results There were 225 cases of NHB of the 500 ABO-HND children in the modeling set with an incidence of 45%. Unifactorial analysis showed that influencing factors were maternal age， premature rupture of membranes， pregnancy with hypertension， diabetes， thyroid disease and anemia， intrauterine distress， cranial hematoma， infection， gestational age， neonatal asphyxia， feeding patterns， low birth weight， intracranial hemorrhage， and respiratory distress（P < 0.05）. Multivariate Logistic regression analysis showed that the independent risk factors were premature birth， neonatal asphyxia， infection， G-6-PD deficiency，intracranial hemorrhage and respiratory distress（OR = 2.377， 7.457， 54.438， 45.487， 2.788，5.407， all P < 0.05）. Three models of bilirubin levels in umbilical cord blood，clinical risk factors and bilirubin levels in umbilical cord blood combined with clinical risk factors predicted the occurrence risk of NHB in ABO-HDN. And the area AUC values under ROC curve were 0.817（95%CI： 0.776 ~ 0.859）， 0.876（95%CI： 0.845 ~ 0.907） and 0.944（95%CI： 0.921 ~ 0.966）， respectively， all P < 0.05. The prediction model of bilirubin levels in umbilical cord blood combined with clinical risk factors was used to verify the practical efficacy of 100 ABO-HDN children in the verification set. The area AUC value under ROC curve was 0.951（95%CI： 0.776 ~ 0.859），P < 0.05. Conclusion The prediction model of bilirubin levels in umbilical cord blood combined with clinical risk factors can ccurately predicted the occurrence risk of NHB in ABO-HDN. It could provide objective reference for prevention， diagnosis， treatment and standardized management of NHB in ABO-HDN.