OHDSI Home | Forums | Wiki | Github

Implementation of Offline Cohort Generation in ATLAS

Dear all,

I wonder if you are aware of this work done by the JnJ team in Belgium (@mvspeybr,@larsh), and others, presented at the European OHDSI symposium (poster link).

See also this nice video demonstrating this functionality:

This allows the distribution of cohorts to others in a network (export and import functionality). It allows every node to be the central node actually which is nice.

I think it is import these efforts are shared to avoid many people are working on the same things.

For example, to what extend can Arachne already do this out of the box? I see additional value in having it in ATLAS as well as shown here?! I think this functionality is very useful and needed and may therefore be on multiple developer’s to do list. I personnally, like the better import and export buttons a lot (some may not even know this is already possible by “miss-using” the export button…).

What does the ATLAS developers team think about this? Can you consider a pull request for this @mvspeybr to the WebAPI and ATLAS github?

Also interested in the Achilles exchange workflow that is mentioned at the end of the poster.

Peter

2 Likes

@Rijnbeek Nice video (and cool music!) I think there is a place for both Arachne, with very rich functionality for collaborative research, and this type lighter weight functionality embedded in Atlas. There has been some discussion on universal identifiers for cohort definitions, which is somewhat related to the export / import process.

I would say that the pull request should go through review by the dev / architecture team. But overall I like it.

Jon

@Rijnbeek - very cool video indeed! As well as the concept

@jon_duke - I agree with you, there are both place and need for ARACHNE and ATLAS.

This is what is available in ARACHNE today:

  • Ability to import any ATLAS design artifact from ATLAS instances connected to site’s ARACHNE Data node. Imagine multiple sites (Columbia, Georgia Tech, IQVIA etc…) are participating in a single study and able to design cohorts and analyses locally and then researches can import and execute it via ARACHNE across distributed data sets available in that study

  • Ability to package and submit imported ATLAS artifacts as a self-contained bundle (packrat for R analyses)

  • Store analysis design, code and results as a single results package in JCR compliant results repo, with quick visualization available

To dream about the bright future a little, here are some ideas that have been floating around and I would like to see in place sometime soon:

  • Create a single global repo for various design artifacts - concept sets, cohorts, analysis etc… Then ATLAS can publish re-usable artifacts and both ARACHNE and ATLAS can read from that repo - not just importing a JSON file. Possibly, we can extend ATHENA to become a platform for distribution of OMOP vocabularies, design artifacts and even final study papers/insights?

  • This could only be possible if we standardize on concept set, cohort and analysis design definitions (JSON?) similarly to what OHDSI have done with OMOP CDM and OMOP Standard Vocabs. This would also enable other vendors to adopt it and be OHDSI compliant

1 Like
t