Md. Rokon Islam Emon

SAGE: Stability-Aware Graph-Based Ensemble Feature Selection for Explainable Postpartum Depression Risk Prediction

SAGE combines information-theoretic relevance, PCA-based structure, and graph-based interactions with bootstrap stability weighting to identify robust and non-redundant predictors of postpartum depression, and allows instance-based insight into selected features from the graph, enabling personalized risk assessment.