OHDSI Home | Forums | Wiki | Github

Achilles in Review


(Frank DeFalco) #1

I’ll be presenting a review of the Achilles package and AchillesWeb on an upcoming OHDSI community meeting. I would like to include some utilization statistics or community reviews as part of that presentation.

Do you and/or your organization currently use Achilles? If so…

How often is it run on your data?

What are the primary insights you use Achilles to obtain?

Is there something you would like to share about your experiences with Achilles?

Is there a particular need you would like to see Achilles meet in the future?


(Don Torok) #2

We create OMOP CDMs for a number of customers. We run Achilles for every ETL. We delivery the Achilles Heel report (along with DQ Dashboard results) with the CDM. There are a number of Achilles Heel tests that, because of missing information in the source data or customer requirements, consistently result in Heel Errors and Warnings. But even for these ‘Acceptable errors’ we compare the counts from one release to next because a spike in the count can indicate a new problem with the source data or the ETL.

A future Heel enhancement could be to compare the Heel results to a prior Heel report for the same source data and flag big differences in Heel report counts.

(Katy Sadowski) #3

We just recently set up ACHILLES (the version in ATLAS) and are still scoping out how best to utilize it. But the likely plan will be to have it run each time we add new/updated data to our CDM, and use the reports to sense check the CDM contents. Also likely that we’ll set up processes to compare the results run over run as well (but a feature to enable comparisons within the app, as suggested above, would be wonderful :slightly_smiling_face:).