A Comparative Assessment of Synthetic Data Augmentation Methodologies (SMOTE, Variational Autoencoders, and GANs) in Resolving Class Imbalance for Predictive Diabetes Analytics. (2026). The Science Post, 2(2). https://www.thesciencepostjournal.com/index.php/tsp/article/view/168