Utilizing Open-Source Large Language Models to Extract Genitourinary Symptoms from Clinical Notes.

The findings suggest that the best performance was achieved when the prompt included a predefined list of definitions of corresponding ICD-10 codes and restricted the model from making assumptions.