gowthamrao
commented
2 years ago Well - we did make cohort diagnostics only run on cohorts that yielded cohorts (that was the initial design)
I did change that behavior in a develop branch once - for such uses cases
@azimov i think its reasonable to support such use cases. In fact - we can have scenarios where the cohort was never instantiated (i.e. cohort generator was never run) - and we just get design diagnostics back
Thoughts for version 3?
Is there a way to look for orphan concepts for a phenotype where a phenotype does not produce a count.
motivating example: Anaphylaxis in IPCI came up with 0 people: https://data.ohdsi.org/Covid19SubjectsAesiIncidenceRate/ (1) Cohort Counts (2) DB = IPCI (3) Cohorts = C8
C8 = https://atlas.ohdsi.org/#/cohortdefinition/349
There are no entries for this cohort/db on the Orphan Concepts tab. However, their database does contain "orphan" terms. They use 42536383-"anaphylaxis shock", which was not part of our definition but now we know should be. Cohort Diagnostics would have caught this if there had just been 1 person who met the original C8 definition.
Now it could be argued that this can only be seen in Phoebe, however, I think it is work thinking if it is possible to expose in CohortDiagnosis as well.
Thx to @marcel1334 and @azzashoaibi for helping identify this.