Explainable AI for Knowledge Graph-Based Drug Repurposing: Methods, Challenges, and Interpretability Frameworks

This paper critically investigates XAI techniques for knowledge graph-based drug repurposing. It organizes graph feature attribution, path reasoning, counterfactuals, and rule-based logic into a stakeholder-aware taxonomy, examining how transparency can coexist with predictive accuracy in biomedical settings. The work highlights evaluation gaps, the need for audience-specific explanations, and practical considerations for deploying trustworthy AI systems in knowledge-rich drug discovery pipelines.