Research demonstrating how Large Language Models can identify diagnoses not explicitly documented in Electronic Health Records by analyzing clinical notes.

• Addresses critical gaps in EHRs where medications lack links to diagnoses
• Leverages discharge summaries which contain more complete clinical information
• Enables more accurate diagnosis inference, especially during early patient visits
• Improves clinical decision-making and research capabilities

This breakthrough matters for healthcare providers by reducing diagnostic delays, improving treatment selection, and creating more complete patient records without additional documentation burden.

Leveraging LLMs for Predicting Unknown Diagnoses from Clinical Notes