VeriFact combines retrieval-augmented generation with LLM-as-a-Judge architecture to verify factual accuracy of AI-generated clinical text against electronic health records.

• Addresses critical need for factual verification in medical AI applications
• Introduces VeriFact-BHC, a new dataset for evaluating factual accuracy in hospital discharge narratives
• Helps prevent potentially dangerous misinformation in clinical documentation
• Represents an important security layer for responsible AI deployment in healthcare

This innovation matters because medical misinformation can directly impact patient care and safety, while providing a framework for maintaining trust in AI-generated clinical documentation.

VeriFact: Verifying Facts in LLM-Generated Clinical Text with Electronic Health Records