DEEP LEARNING MODEL FOR OPTIMIZING RISK OF MATERNAL MORTALITY RATE ON EARLY PREGNANCY IN NIGERIA

ABSTRACT

Nigeria faces a critical maternal health crisis, with a maternal mortality rate of 512 deaths per 100,000 live births - among the world's highest. This dreadful situation stems from inadequate healthcare infrastructure, limited access to skilled care, and socioeconomic barriers. While several studies have employed machine learning techniques like Random Forest and SVM for maternal health prediction, few have explored deep learning's potential. This study bridges this gap by developing an advanced deep-learning model using LSTM networks and MLP architectures to predict early pregnancy risks. This study hopes to incorporate comprehensive demographic, clinical, and socioeconomic data, utilizing innovative techniques including SMOTE-ENN for data balancing, RFE for feature selection, and K-Fold cross-validation. By integrating explainable AI components, the study will enhance clinical interpretability and practitioner trust. This research provides a scalable, data-driven solution tailored to Nigeria's unique challenges and it will offer a replicable framework for similar low-resource settings. The study will demonstrate deep learning's superior capability in maternal risk prediction compared to conventional machine learning approaches, representing a significant advancement in AI applications for maternal healthcare

Keywords: