OHDSI Home | Forums | Wiki | Github

Atlas Import - Why concept counts are different when imported using json?

estimationmethods
vocabularies
patientprediction
atlas
cdm

(Selva) #1

Hello Everyone,

Q1) I am trying to import a concept set from a public version of Atlas to my local Atlas instance.

Source concept sets can be found here - click here for public Atlas cohort

From this cohort, am copying the json of T1DM Medications concept set.

As you can see that it has 3078 concepts (included concepts) but when I copy the json and put it in my local Atlas instance, I only see 2978 concepts.

May I know what is the issue here? Is this an issue with Atlas import or I have to make some changes in local instance or do any selection further?

This happens for other concepts like T2DM Medications from the same cohort as well.

Q2) When I import a cohort definition json, I don’t have to create concept sets on my own. Am I right? The imported cohort definition (json file) will create the concept sets for us (but makes it available only to the cohort and not to the repository). Am I right?

Can you help me with these two questions?


(Kristin Kostka, MPH) #2

Selva, don’t blame our friend Atlas. :slight_smile: I would guess this has to do with the data under your Atlas and the vocabulary reference tables you’re using to build that CDM. In my view, you have a few things here at play:

  1. The vocabulary version you built your data on.
  2. If you have multiple CDMs, in Atlas, you are then picking which schema you’re pulling the vocabulary reference tables from.

If this sounds confusing, let’s go to the Configuration tab. In Public Atlas, you can see the concept sets are pulling from:

The version of vocabulary that you use can, in my experience, change the counts you get.

Your other question:

Correct, you’re bringing over the Cohort Definition and the underlying concept sets – whether they are used in the cohort definition are simply stored in the cohort definition unused. You’ll have to click that green button at the bottom to push the concept sets to your local repository.


(Selva) #3

Thank you @krfeeney. Will try updating the vocabulary version. Our vocabulary version is V5.0 10 JAN -19 And yes, Atlas has been a very useful tool for us to experiment with things and learn… I overlooked the vocabulary version update issue at our end. My bad.


(Selva) #4

Hi @krfeeney,

As I am unable to upload the recent vocab (Jan 2020) files due certain issues listed here , May I check with you on whether the concept count difference can also be due to any other issues?

If yes then while I wait for vocab file issues to be sorted, I can verify whether everything else w.r.t to concept count is right


(Kristin Kostka, MPH) #5

@SELVA_MUTHU_KUMARAN, my friend, this is where my ATLAS know-how hits its edge. I only know what I’ve experienced. I don’t debug vocabulary files – someone in my ETL shop loads and maintains my reference tables.

I would have to defer to other folks like @Manan @Konstantin_Yaroshove @Eldar @Chris_Knoll.


(Akshay Kumar) #6

Yes, selecting the right schema resolved the issue for us. Thank you @krfeeney


t