Related Experiment Video
Updated: Apr 2, 2026
Analysis of Congenital Heart Defects in Mouse Embryos Using Qualitative and Quantitative Histological Methods
Published on: March 10, 2020
Explainable machine learning model for identifying key risk factors in congenital heart disease prediction using
Yunyan Wang1,2, Tiezheng Li2, Jiajia Chen2
1Heart Center, Children's Hospital, Zhejiang University School of Medicine, Zhejiang Key Laboratory of Neonatal Diseases, National Clinical Research Center for Children and Adolescent's Health and Diseases, Hangzhou, 310052, China.
This study developed a cost-effective machine learning model to predict congenital heart disease (CHD) in newborns, addressing screening challenges in low-resource areas. The model identified key risk factors, aiding early detection and improving clinical decision-making for neonatal CHD.
Area of Science:
- Neonatal Health
- Cardiology
- Machine Learning in Healthcare
Background:
- Neonatal congenital heart disease (CHD) screening faces challenges in low-income regions due to physician shortages.
- There is a need for cost-effective methods to identify high-risk infants for CHD screening.
Purpose of the Study:
- To develop and validate a non-invasive, cost-effective prediction model for neonatal CHD.
- To identify key risk factors associated with CHD in newborns.
Main Methods:
- A multicenter case-control study involving 1,633 mother-infant pairs.
- Utilized machine learning algorithms (XGBoost, SVM, RF, LightGBM, LR) and LASSO regression for model development.
- Employed SHAP and logistic regression to identify significant risk factors.
Main Results:
- The XGBoost model achieved the highest AUC (0.724 internal, 0.706 external validation).
- Key risk factors identified include COVID-19 infection pre-conception, gestational medication use, and viral infections during pregnancy.
- Identified risk factors showed significant odds ratios, indicating increased CHD risk.
Conclusions:
- The developed CHD prediction model is non-invasive, cost-effective, and performs robustly.
- This model can supplement routine screening, aiding in identifying high-risk newborns.
- The machine learning approach can enhance clinical risk assessment and decision-making for neonatal CHD.
