I really like the idea of sharing computable definitions. Will the up and
coming OHDSI phenotype group work to build the features that @mark_danese
mentioned as well as validation specs? Will this be a like a computable
PheKB?
1 Like
Patrick just scared the software straight. I think maybe it was the operator error. 
By the way, I wonder what people are doing, thinking of doing, planning on doing with the NLP Note table?
I have developed a doc2vec model (based on the NIH 20+ million abstracts) that would allow one, based on raw clinical note text, to locate related NIH abstracts (and therefore full articles, if available) related to the subject matter.
Depending on how often the model is updated (with new NIH abstracts) one can quickly locate the most recent research related to a specific patient.
Well, this thread appears to have died from lack of treatment. Does anyone know how the SQL being generated from the UI was tested for correctness?
Aren’t there a bunch of cohort definitions that are pre-defined to yield specific sets of person_id(s) for the testing purposes? How would anyone, who installs the software in his environment, verify that all is well?
Thanks.