Treffer: Evaluating Advances in Machine Learning Algorithms for Predicting and Preventing Maternal and Foetal Mortality in Nigerian Healthcare: A Systematic Approach

This study systematically analysed developments in machine learning (ML)-based prediction algorithms aimed at reducing maternal and foetal mortality in Nigerian hospitals. Key causes of maternal death in Nigeria include obstetric haemorrhage, eclampsia, sepsis, obstructed labour, and complications from unsafe abortions. The comparison of maternal mortality ratios between Nigeria and developed countries highlights significant disparities, emphasizing the need for targeted interventions. This research employed a random-effects model to synthesize effect sizes from multiple studies, accounting for variations in study populations and hospital settings. Metrics such as precision, accuracy, recall, and F1-score were used to evaluate ML algorithms including logistic regression, decision trees, random forests, support vector machines (SVMs), neural networks, and ensemble methods. The results indicate high prediction accuracy of 80-90% for these algorithms, with neural networks performing best at 90% accuracy. The implementation challenges such as data quality, limited technology access, and ethical considerations pose significant barriers. Improving data infrastructure, fostering interdisciplinary collaboration, and establishing ethical frameworks are crucial for successful ML integration in healthcare. The study emphasized on the potential of ML to transform maternal and foetal healthcare through early detection, personalized care, and optimized resource allocation, with more emphasis on the need for holistic approaches to address both technical and socio-cultural challenges in Nigeria. Future research should focus on developing robust ML algorithms, enhancing data interoperability, and promoting a data-driven culture to improve maternal health outcomes.