SEP 4, 2026 · PREPRINT
A Comparative Study of Counterfactual Explainers for Graph Neural Networks Enabling Multiple Types of Graph Edit
arXiv
A comparative evaluation study of existing explainability methods without novel intervention, clinical outcome, or definitive evidence—raises questions about method performance rather than answering clinical or translational questions.
Study details
DesignComparative methods evaluation study
InterventionSix state-of-the-art counterfactual explain…
ComparatorMethods compared against each other using q…