先天性心脏病患儿术后心脏骤停风险预测模型的构建和验证  

Development and Validation of a Risk Prediction Model for Sudden Cardiac Arrest in Children With Congenital Heart Disease After Surgery

作　　者：刘亚飞 邢海英 张茜 冯沃蕾 祝芳菲 王艳娇 刘世琼 马艳 LIU Yafei;XING Haiying;ZHANG Qian;FENG Wolei;ZHU Fangfei;WANG Yanjiao;LIU Shiqiong;MA Yan(Pediatric Surgery Recovery Room,National Center for Cardiovascular Diseases and Fuwai Hospital,Chinese Academy of Medical Sciences and Peking Union Medical College,Beijing 100037,China;Operating Room,National Center for Cardiovascular Diseases and Fuwai Hospital,Chinese Academy of Medical Sciences and Peking Union Medical College,Beijing 100037,China)

机构地区：[1]中国医学科学院、北京协和医学院、国家心血管病中心、阜外医院小儿外科恢复室,北京100037 [2]中国医学科学院、北京协和医学院、国家心血管病中心、阜外医院手术室,北京100037

出　　处：《中国循环杂志》2025年第3期254-260,共7页Chinese Circulation Journal

基　　金：中国医学科学院阜外医院护理部专项科研基金(HLB2020001)。

摘　　要：目的:构建先天性心脏病(先心病)患儿术后心脏骤停风险预测模型并评价其预测效能,为心脏骤停的预防、病情评估提供参考。方法:回顾性收集2020年1月1日至2022年5月31日于中国医学科学院阜外医院住院的5029例先心病术后患儿病历资料,根据术后是否发生心脏骤停分为心脏骤停组(n=33)和未发生心脏骤停组(n=4996)。采用R软件在训练数据集上建立术后心脏骤停风险预测的随机森林模型,在验证数据集上采用预测准确度、灵敏度、特异度、阳性预测值、阴性预测值等评价模型的预测效果。结果:心脏骤停发生率为0.66%,生存率为72.73%。采用随机森林算法通过变量重要性评分对先心病术后发生心脏骤停危险因素的重要程度进行排序,得到排名前6位的重要预测变量为:术后第1天的血压、乳酸水平、心律、心率、动脉氧分压、血氧饱和度。通过随机森林算法对训练集建立的模型在测试集上验证,得出预测准确度为99.8%、特异度为87.5%、灵敏度为99.9%、Kappa值为0.8225、阳性预测值为99.9%、阴性预测值为77.8%。结论:本研究构建的先心病患儿术后心脏骤停风险预测模型,预测效能良好,有助于医护人员及早干预,避免心脏骤停发生。Objectives:To develop a risk prediction model for sudden cardiac arrest(CA)in children with congenital heart disease(CHD)after surgery and validate its predictive efficacy,providing a reference for the prevention of CA and risk stratification.Methods:Medical records were retrospectively analyzed from 5029 children who were hospitalized in Fuwai Hospital,Chinese Academy of Medical Sciences from January 1,2020 to May 31,2022 and underwent CHD surgery.The patients were divided into two groups:those who experienced CA after surgery(n=33)and those who did not(n=4996).A random forest model for predicting the risk of postoperative CA was established on the training dataset using R software,and the predictive effect of the model was evaluated on the validation dataset using indicators of predictive accuracy,sensitivity,specificity,positive predictive value,negative predictive value.Results:The incidence of CA in this center was 0.66%,survival rate is 72.73%.Using the random forest algorithm,the importance of risk factors for sudden CA after CHD surgery was ranked by variable importance scoring,with the following top 6 important predictive variables:blood pressure,lactate levels,heart rate,cardiac rhythm,arterial oxygen partial pressure,and blood oxygen saturation on the first day after surgery.The model established by the random forest algorithm on the training set was validated on the test set,yielding a predictive accuracy of 99.8%,specificity of 87.5%,sensitivity of 99.9%,kappa coefficient of 0.8225,positive predictive value of 99.9%,and negative predictive value of 77.8%.Conclusions:The established prediction model of sudden CA in children with CHD after surgery had good performance.It might help medical staff on decision making of early intervention,preventing the occurrence of CA,and improving the outcomes of children with high risk of CA post surgery.

关 键 词：先天性心脏病 心脏骤停 危险因素 随机森林 预测模型 

分 类 号：R54[医药卫生—心血管疾病]