GEMeX introduces a comprehensive benchmark dataset that enables medical AI systems to not only answer questions about chest X-rays but also explain their reasoning.

• Addresses critical gaps in current Med-VQA datasets by including visual and textual explanations for diagnostic answers
• Expands question diversity to better reflect real-world clinical scenarios beyond limited formats
• Enhances medical education by providing explanations that improve understanding for patients and junior doctors
• Advances interpretability in AI-assisted medical diagnosis, a crucial factor for clinical adoption

This research represents a significant step toward more transparent, trustworthy AI systems in healthcare settings where explainability is essential for both regulatory approval and physician confidence.

GEMeX: A Large-Scale, Groundable, and Explainable Medical VQA Benchmark for Chest X-ray Diagnosis