Document Type
Article
Publication Date
Winter 12-28-2025
Abstract
Congenital heart defects (CHD) are heart malformations present at birth, affecting heart function and circulation, and are a leading cause of infant mortality. CHD can result from genetic, environmental, and maternal health factors, making early detection essential. Early diagnosis allows for timely intervention, reducing risks like heart failure or stroke. In countries like Egypt, CHD often remains undiagnosed due to limited healthcare resources. Artificial intelligence (AI) can improve early detection by analyzing risk factors. This study presents a predictive model for CHD using maternal and paternal health factors. Data was collected from 571 families: 260 with a CHD-affected child and 311 with healthy children. After preprocessing the data, ten machine learning models were tested, including Random Forest (RF), Decision Tree (DT), and MLP Classifier. RF achieved the highest accuracy at 97.37%, followed by DT at 96.49%, and MLP at 92.96%. The results show AI’s potential in predicting CHD, supporting early diagnosis and improving infant outcomes.
Recommended Citation
Abdo, Amany M. Prof.; Mosallam, Asmaa M. Ms.; and Abdelhamid, Laila M. Assoc.Prof., "Using Machine Learning to Predict Women at Risk Having a Child With Congenital Heart Defects" (2025). Information Systems. 19.
https://buescholar.bue.edu.eg/info_sys/19
Comments
ML has demonstrated significant potential in predicting CHD, aiding early diagnosis and intervention. This study leveraged a balanced dataset to evaluate ten ML models, with the RF model achieving the highest performance (accuracy: 97.37%, F1-score: 96.77%, and AUC: 1). Unlike prior research, our approach incorporated a broader range of risk factors, addressing key limitations, such as imbalanced data and restricted feature sets. The study underscores the strength of ML in handling complex, non-linear relationships, providing both practical and theoretical justification for RF’s effectiveness. Future directions should focus on integrating additional risk factors, such as genetic predispositions and environmental influences, to enhance predictive reliability. This research sets a foundation for more advanced artificial intelligence-driven methodologies in CHD prediction, emphasizing the role of ML in transforming healthcare analytics.