#artificial intelligence
Apr 2026
Iterative Multimodal Retrieval-Augmented Generation for Medical Question Answering
MED-VRAG is proposed, an iterative multimodal RAG framework that retrieves and reasons over PMC document page images instead of OCR'd text, and reaches 78.6% average accuracy across four medical QA benchmarks.
Xu-Peng Chen, Binbin Shi, Chen-Qian Le et al.
· arXiv.org · 1 citation