Predicting Maternal Health Risks Using Neural Networks and Feature Engineering
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
A total of 1,014 maternal patients from a clinical setting developed a modeling framework for estimating maternal health risk incorporating physiological indicators of age, temperature, blood glucose, heart rate and blood pressure. Clinical indicators were distilled by advanced medical feature engineering to discover meaningful indicators such as blood pressure ratios, pulse pressure, mean arterial pressure, and a risk score for overall assessment. Maternal risk was examined and classified as low, mid, and high risk using ensemble machine learning methods (XGBoost, LightGBM, Random Forest and Gradient Boosting) and a modified neural network. Class imbalance was identified and addressed with the ADASYN approach, while an explainable AI approach using SHAP was applied to facilitate interpretation of the outcomes from both ensemble and deep learning models. The highest-performing method yielded 95% accuracy and F1 score of 0.95, suggesting that combining clinical feature engineering with ensemble and neural network methods of modeling maternal health risk may be effective.
Fields of Study & Subjects
Artificial Intelligence in Healthcare
Machine Learning in Healthcare
Imbalanced Data Classification Techniques
Publication details
- Journal / source
- International Conference on Intelligent Systems for Communication, IoT and Security (ICISCoIS)
- Pages
- 1–7
- Publication date
- 8 January 2026
- Language
- EN