Relationships between entities with identical meanings
@blaze999 Thank you for your work on the model - it's amazing! Probably the best clinical NER I've seen on HF (if you know otherwise, I'd appreciate a pointer in the right direction). My task is to find values for certain labs or procedures in clinical documents, but I'm struggling with relating abbreviations and differently phrased terms which have identical meanings to be recognized ass exactly the same token. Example: I want to make sure that EF, EF%, %EF, LVEF, VEF, "ejection fraction", "fractional ejection", etc. - all mean the same thing and are captured as their higher level equivalent (e.g. "EF" should be treated as "ejection fraction"). I really don't want to create a manual ontology. Seems to defy the purpose of using LLMs. Any recommendations? Also, how would you go about getting the value associated with the measurement. In this case, EF/VEF/etc is captured as a DIAGNOSTIC_PROCEDURE token, and values (text and numeric, such as "normal", "reduced", ">50%" etc.) are detected as LAB_VALUE. >50% is actually detected as a series of two named entities.
Thanks in advance! Really appreciate your effort and your feedback!