The significance of maternal health cannot be overemphasized, and the ability to predict maternal outcomes accurately is critical to ensuring the well-being of both mothers and infants. This study presents ConvXGB, a novel predictive model that utilizes a combination of XGBoost, a potent gradient boosting algorithm, and deep learning to extract intricate features. The objective is to enhance precision and robustness of maternal outcome predictions. The study sourced diverse maternal health data from the southern region of Nigeria and implemented Synthetic Minority Over-sampling Technique (SMOTE) to address any dataset imbalances. Results obtain demonstrate a significant improvement in model performance, with an accuracy rate of 97.96% across various maternal outcome classes. The recommendations from this study highlight the potential of ConvXGB in advancing maternal health predictive analytics, supporting informed clinical decision-making, and improving resource allocation. Further studies are warranted to explore the broader applicability of ConvXGB in different healthcare domains through outcome analyses and methodological advancements.