Development and Validation of a Risk Prediction Model for Sudden Cardiac Arrest in Children With Congenital Heart Disease After Surgery
10.3969/j.issn.1000-3614.2025.03.008
- VernacularTitle: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
- Author:
Yafei LIU
1
;
Haiying XING
;
Qian ZHANG
;
Wolei FENG
;
Fangfei ZHU
;
Yanjiao WANG
;
Shiqiong LIU
;
Yan MA
Author Information
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
- Publication Type:Journal Article
- Keywords:
congenital heart disease;
cardiac arrest;
risk factor;
random forest;
prediction model
- From:
Chinese Circulation Journal
2025;40(3):254-260
- CountryChina
- Language:Chinese
-
Abstract:
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 5 029 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=4 996).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 staffon decision making of early intervention,preventing the occurrence of CA,and improving the outcomes of children with high risk of CA post surgery.
