利用分类树模型构建危重新生儿预后预警模型  

Construction of a prognostic early warning model for critically ill neonates using a classification tree model

作　　者：陈云杰 胡忠华 梁晓燕 戚利那 Chen Yunjie;Hu Zhonghua;Liang Xiaoyan;Qi Lina(Linping Campus,the Second Affiliated Hospital Zhejiang University School of Medicine,Hangzhou Zhejiang 311100,China)

机构地区：[1]浙江大学医学院附属第二医院临平院区,浙江杭州311100

出　　处：《护理与康复》2025年第1期9-14,共6页Journal of Nursing and Rehabilitation

基　　金：杭州市医药卫生科技项目,编号B20230024。

摘　　要：目的利用分类树模型构建危重新生儿预后预警模型,帮助临床医护人员在复杂的临床情境中做出更加科学、合理的判断和决策,并采取有效的防治措施降低危重新生儿死亡风险。方法选取2021年1月至2023年7月浙江大学医学院附属第二医院临平院区儿科收治的143例危重新生儿为研究对象,根据当次住院结局分为预后良好组(n=105)和预后不良组(n=38)。比较两组患儿的临床资料,应用分类树模型构建危重新生儿预后预警模型,应用灵敏度、特异度、预测准确度及受试者工作特征曲线下面积评估模型预测能力。结果预后不良组小于胎龄儿、胎儿窘迫、1 min Apgar评分<4分、5 min Apgar评分<4分、PaO 2<50 mmHg、碱剩余<-7 mmol/L、母亲年龄≥35岁、羊水异常比例均高于预后良好组,差异有统计学意义(P<0.05)。多因素logistic回归分析结果显示,1 min Apgar评分<4分、PaO 2<50 mmHg、碱剩余<-7 mmol/L、小于胎龄儿、胎儿窘迫、母亲年龄≥35岁、羊水异常进入回归方程,差异有统计学意义(P<0.05)。构建的危重新生儿预后预警的分类树模型包括4层、17个节点、9个终末节点,共筛选出7个解释变量,即1 min Apgar评分、PaO 2、碱剩余、小于胎龄儿、胎儿窘迫、母亲年龄、羊水异常,该模型的风险率为0.187,提示该模型拟合效果好。结论分类树模型能有效拟合危重新生儿预后危险因素的预测,对危重新生儿预后具有良好预测价值。Objective To construct a prognostic early warning model for critically ill neonates using a classification tree model,assisting clinical medical staff in making more scientific and rational judgments and decisions in complex clinical situa-tions,and implementing effective prevention and treatment measures to reduce the risk of neonatal mortality.Methods A total of 143 critically ill neonates admitted to the pediatrics department of Linping Campus,the Second Affiliated Hospital Zhejiang University School of Medicine between January 2021 and July 2023 were selected as study subjects.Based on hospitalization outcomes,they were divided into a good prognosis group(n=105)and a poor prognosis group(n=38).The clini-cal data of the two groups were compared,and a classification tree model was used to construct a prognostic early warning model for critically ill neonates.The model's predictive ability was evaluated using sensitivity,specificity,predictive accuracy,and the area under the receiver operating characteristic curve.Results The proportions of small for gestational age infant,fetal distress,1-minute Apgar score<4,5-minute Apgar score<4,PaO 2<50 mmHg,base excess<-7 mmol/L,maternal age≥35 years,and abnormal amniotic fluid were significantly higher in the poor prognosis group than in the good prognosis group(P<0.05).Multivariate logistic regression analysis indicated that 1-minute Apgar score<4,PaO 2<50 mmHg,base excess<-7 mmol/L,small for gestational age infant,fetal distress,maternal age≥35 years,and abnormal amniotic fluid were significant factors included in the regression equation(P<0.05).The constructed classification tree model for the prognostic early warning of critically ill neonates comprised 4 levels,17 nodes,and 9 terminal nodes,identi-fying 7 explanatory variables,which were 1-minute Apgar score,PaO 2,base excess,small for gestational age infant,fetal distress,maternal age,and abnormal amniotic fluid.The model's risk rate was 0.187,indicating good model fit.Conclusion The classification tree model ca

关 键 词：危重症 新生儿 预后 危险因素 分类树 预警模型 

分 类 号：R473.72[医药卫生—护理学]