A Robust Machine Learning Predictive Model for Maternal Health Risk

Abstract

Maternal health refers to the physical, mental, and emotional well-being of mothers throughout all stages of pregnancy, birth, and the postpartum period. Morbidity and death rates during pregnancy are significant health statistics because they provide information about the accessibility of maternal and other medical resources. The main reason of maternal mortality are hemorrhage, high blood pressure, early labor, and abortions that are performed under dangerous conditions. Machine learning algorithms play a significant role in determining maternity health risks. In this paper, Traditional Machine Learning algorithms are applied for maternal health risk prediction. Model performance has potential for improvement, thus a more robust and dependable machine learning model has been proposed to operate even in worst, average, and best scenarios and provide the robust performance by considering the performance of all scenarios. The proposed robust model turns out to be the most efficient robust model among all with an accuracy of 70.21%, which is quite satisfactory when compared to traditionally applied ML Models.