AIMEN: AI Framework for Predicting Neonatal Health Risks

A team of researchers has introduced a deep learning framework called Artificial Intelligence for Modeling and Explaining Neonatal Health (AIMEN), designed to forecast negative labor outcomes based on maternal, fetal, obstetrical, and intrapartum variables. AIMEN enhances interpretability by utilizing what-if scenarios, illustrating how changes in input variables can impact predicted results. To tackle issues of class imbalance and limited sample sizes, the framework incorporates Conditional Tabular GAN (CTGAN) for data augmentation, which includes generating synthetic data and adjusting feature bounds for certain training samples. This initiative seeks to bridge the gap in reliable automated systems for clinical decision-making during childbirth, aiming to facilitate early interventions to avert or lessen adverse effects like cerebral palsy.