This study demonstrates that Retrieval-Augmented Generation (RAG) significantly improves Large Language Models' performance in pancreatic cancer staging by providing access to reliable external knowledge.

• RAG-enhanced models (NotebookLM) achieved superior accuracy in staging pancreatic cancer compared to standard LLMs
• Models with RAG showed 32.4% higher accuracy in T-staging and 27.0% higher accuracy in overall staging
• RAG capabilities reduced hallucinations and improved clinical reliability in complex medical assessments
• Findings suggest RAG is essential for medical applications where precision and factual accuracy are critical

This research matters for healthcare by offering a pathway to more reliable AI-assisted diagnosis tools that maintain accuracy while providing explainable results, potentially improving clinical decision-making in oncology.

Enhancing Pancreatic Cancer Staging with Large Language Models: The Role of Retrieval-Augmented Generation