Abstract
Acute hemorrhage in pregnancy may lead to maternal and/or fetal morbidity or mortality. In emergency medicine, blockage of the aorta via an inflatable endovascular balloon, technically referred to Resuscitative Endovascular Balloon Occlusion of the Aorta (REBOA), is used to manage hemorrhage. However, the application of REBOA in pregnancy needs to strike a balance between two competing objectives of limiting maternal blood loss and ensuring fetal wellness, for which one would need to predict the impact of regulated blood pressure on fetal wellness. To address this problem, we propose an efficient machine learning-based method to predict the temporal impact of the distal Mean Arterial Blood Pressure (dMAP) controlled by the REBOA on the oxygen content in the fetal blood. Evaluation of the algorithm on data collected from in-vivo experiments from pregnant ewe animal models exhibits mean absolute error of 0.61, 1.09, 1.42, 1.70 mmHg, and coefficient of determination of 0.95, 0.86, 0.76, 0.64 for prediction of partial pressure of oxygen in fetal arterial blood, a key predictor of fetal wellness, in 2.5, 5, 7.5, 10-minute prediction horizons, respectively.