Health-Informatics-UoN / carrot-mapper

Convenient and Reusable Rapid Omop Transformer.
https://carrot.ac.uk
MIT License
14 stars 4 forks source link

Concept reuse conflicts ICD10 codes #902

Open erummas opened 3 weeks ago

erummas commented 3 weeks ago

Is there an existing issue for this?

Current Behavior

Hello, I have observed that the concept reuse is not working fine on the ICD10 codes. sharing screenshots for only two cases Image In the image above the automated concept_id generated is incorrect (red) and only one of the reuse one is correct(blue)

Image In the above image the reuse concept id is incorrect and only the automated one is correct

ScanReport No. 701

Expected Behavior

Show only valid mappings

Steps To Reproduce

No response

Environment

I'm part of a Project Team

No response

Anything else?

No response

Are you willing to contribute to resolve this issue?

None

AndyRae commented 3 weeks ago

Thanks @erummas, we'll investigate.

For image .2 - a quick look is that we had a test Scan Report that implemented that rule, and was being reused, I've turned this off now. But this is obviously a general concern of ours, as currently effectively one Scan Report poisons the well... so some of our upcoming work will prevent this with more control of what mappings are reused from where.

Image .1 might be the same, but I want to check.

erummas commented 3 weeks ago

Thanks @AndyRae , yes my concern with image 1 is the incorrect Automated mapping.

AndyRae commented 3 weeks ago

We're having to investigate broadly now - as we follow this through, it looks as though Carrot is behaving as expected. The data in the vocabularies downloaded from Athena does indicate a non-standard to standard mapping exists here, but Athena itself does not.

erummas commented 3 weeks ago

Hi @AndyRae are you talking about image 1? The image shared below Non standard to standard mapping on Athena Image

For image 2 Image

erummas commented 3 weeks ago

@AndyRae it could be a case of vocab versions. If after the mapper upgrade then that could explain the different mappings. I have checked Athena and it states last SNOMED update 01-Feb-2024

AndyRae commented 3 weeks ago

@erummas versioning is definitely part of what I am trying to understand here.

To focus on image 1 as it's easier for me...

We're mapping this Concept: https://athena.ohdsi.org/search-terms/terms/45755372 Which has a Standard mapping to: Acute alcohol intoxication

But Carrot shows this as a Standard mapping (but Athena doesn't!): https://athena.ohdsi.org/search-terms/terms/36714559

Our current CDM vocabulary shows this non-standard to standard: Image (Maps to is a Non-standard to Standard map (OMOP)) (this is what I mean by Carrot is behaving as I would expect)

Note the validity dates - this was introduced back in February.. I checked (restored on Athena) our raw vocabularies downloaded on 27 March, and 8 October, and this mapping was present in both. And again most recently downloaded on 31 October, where now the mapping is not present. So we can fix this when we update Carrot to use these vocabularies (this is in progress).

Vocabulary updates are something of a mystery to me, as I expected them until now to follow the release cycle, with most recent update 30 August.
I could not find any OHDSI documentation where other updates might be taking place, but this does imply it is to me. On the Athena list there is for example, ICD10CM updated on 01-Oct-2024. Which shows an update taking place outside of the release cycle. But not for ICD10 or SNOMED. I am still investigating a bit to make sure I understand our issue properly, and I haven't missed anything, before I ask OHDSI, so any thoughts you have would be welcome.

I would note - we fixed a bug where multiple non-standard to standard mappings weren't working - it would just take the first one. This has now been fixed and deployed.

erummas commented 3 weeks ago

Hi @AndyRae, you can find the vocabulary release roadmap here https://github.com/OHDSI/Vocabulary-v5.0/wiki/Release-planning I see that we have a meeting set up for tomorrow. We could discuss vocab updates in more detail then.