Unveiling the Deep Learning Techniques for Maternal Health Risk Prediction

Preventing pregnancy complications and improving outcomes for mothers and infants depend on early detection of maternal health risks. The purpose of this study is to assess and compare how well various machine learning classification algorithms predict maternal health risks based on physiological data. The models examined include both classical machine learning classifiers such as Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), K-Nearest Neighbors (KNN), Decision Tree (DT), Bagging, Random Forest (RF) and Support Vector Machine (SVM) and advanced deep learning-based tabular data models specifically TabTransformer and SAINT (Self-Attention and Intersample Attention Transformer). Split validation was used to discrete the observation dataset as training and testing samples and 10-fold cross-validation was applied for additional valuation. F1 Score, recall, precision, and accuracy were employed to assess the performance of the model. The findings show that although conventional models offer reliable baseline predictions TabTransformer and SAINT produce better accuracy and robustness particularly SAINT which consistently outperformed others across all metrics. These results demonstrate the potential of deep learning-based methods for early intervention support and real-time maternal health risk assessment.