The pair of you should contact Jan (in response to Data analysis tool - #2 by janflowers).
But it’s going to be slower than you’d like to get actual deidentified data from an implementation that has expressed interest. I recommend that @raff follows up on that lead, and @lluismf goes ahead and generates some random diagnoses.
I would not approach it this way.
So, the demo data that you’re using is based off an old OpenMRS approach and concept dictionary, and it’s not what we current recommend to people. (This could also explain the concepts without a preferred name situation.)
It would be better to use the reference application + the CIEL dictionary (or just the subset that’s packaged with the refapp) + reference demo data. This will auto-generate diagnoses, with the right structure. In that structure you’d want to look for obs where concept_id = 1284. (technically it should really be obs where obs.concept has a SAME-AS mapping to “org.openmrs.module.emrapi:Coded Diagnosis” but I would ignore this during the spike).