search

Georgetown-IR-Lab / QuickUMLS

System for Medical Concept Extraction and Linking
MIT License
369 stars 95 forks source link

Unable to return match for chron's disease #85

Closed dimitri-kaid closed 1 year ago

dimitri-kaid commented 2 years ago

I am unable to return a cui for Chron's disease, chrons disease, or any variation thereof; I can get a cui for IBD1, regional enteritis, and other code descriptions that share the same CUI (C0010346). I am running python 3.8.5 and quickumls 1.4.0 on Mac 12.2.1. I built the QuickUMLS database (using the -L and -U flags) from a 2020 version of UMLS with 38 unique code descriptions for C0010346.

In core.py, ngram_cands (which seems to invoke def retrieve(self, *args): return _simstring.reader_retrieve(self, *args) in simstring.py) are empty for any permutation of chron's disease; the code returned by the matcher is for disease (C0012634) and for "chron's" no match is returned.

dimitri-kaid commented 1 year ago

And the reason is that the concept is crohn's disease. No issue with the library.