Downloading vocab v20240229 via Athena gives a more recent version of the vocab. We compared it with an old (subselection) downloaded 2 years ago, and gives different content. Also, looking at the vocabulary table, it shows dates in the future (for example SNOMED v2024-04-10 for UK). Download looks like an unreleased/developers vocab version somewhere between 2 official releases.
This old vocab release can still be used by datapartners. Its important for evaluating compatibility of conceptsets to have insight in older vocabulary versions.
Hopefuly the original v20240229 can be restored in Athena.
When you download vocabularies from ATHENA, it gives actual published vocabularies version. You can see the actual version in ATHENA by clicking (?) icon at the top right.
Also you can check vocabulary.csv file in the archive and see the row with vocabulary_id=“None”. It’s a record that contains the version of vocabularies in your zip archive.
Its NOT about downloading the current vocabulary version. Its about about downloading the release of 1,5 years ago: v20240229. Af far as I can see, the zip we get is NOT the correct version and also does not look like any of the published vocabulary versions.It looks like there is something with the v20240229 version thats currently in Athena.
By the way, the “None” vocabulary indicates that is is v20240229, but as far as we can see this is NOT that version. See also, the SNOMED record in this table. This has a future date in the version text. Some other records the same. We have a subset of the vocab release downloaded long time ago and the concept table has other set of standard/non-standard flags (this triggered the difference).
The vocabulary release process in ATHENA takes place twice per year, and no changes are made to vocabularies in between. There is no technical option for unreleased or developer versions.
For example, SNOMED v2024-04-10 for the UK was released in v20250227.
Therefore, it is difficult to determine exactly which file you have and how it was prepared.
If you have concerns, please open a ticket on GitHub.
The dates of individual vocabs can indeed be in future based on what the source gives us. Maybe Marcel can give some examples of concepts that differ among his version and Athena version.
@mdewilde I thought you asked me about whether we have earlier versions (even pre-2024) in the email and I did reply to you. Didn’t it get through?
If that selection box in screenshot that Anna send, that is even named “choose your version” is NOT for downloading from previous version, please explain what this box is for…
It seems to work on the other versions…
ConceptId that triggered this whole thing was conceptid=37392365. Standard concept in v20240229 downloaded long time ago (few days after the release). But suddenly a NON-Standard concept when re-downloaded v20240229 recently. Vocabulary table with version attributes of some vocabs showed that is was a more recent version, even that the None entry says its v20240229. Versions after this, the overlapping records of vocabulary table of old downloads and new downloads are the same (as expected). So far, only for v20240229 there is something weird.
@aostropolets : yes you email got through, sorry I’ve not found the time to respond yet. Also, partly becuase I’m not sure (yet) if the vocab from 2022 is useful for our goal since this is before the “more official” half-year releases. First half-year version was end 2023 if I’m correct. So, if you can find a “full” vocab release of this version, AND I’m able to get my hands on a correct “full” version of v20240229, it would mean we have “all” half-year releases until now (would be 5 versions). That would be super to get insight in the effect on conceptset definitions on these “old” vocabs and also the effect of old definitions on later vocab versions. All I need is a consisitent chain of “full” vocob versions, ideally from end 2023 until now. Based on some old downloaded (where we only had a subselection of vocabs selected), we already see some interesting things to be aware of when designing conceptsets to work on multiple vocab versions (what you cannot prevent in multi-center studies), or using an old definition on a later vocab release.
Looking forward to seeing you face-to-face and discuss further next week.
@aostropolets@mdewilde Sorry for confusion. Yes, there are historical versions since 2024 releases. I was under impression that it’s available for diff only. I corrected my answers to avoid confusion for others.
@Konstantin_Yaroshove: No problem. Very usefull for reproducability and for checking compatibility between various vocab versions. Hopefuly old vocab versions will be re-downloadable for several years from now.
@aostropolets: Yes, correct for this conceptid example. And the content of the vocabulary table also shows some vocabs with versions after 2024-02-29 for example SNOMED version attribute:
“2023-07-31 SNOMED CT International Edition; 2023-09-01 SNOMED CT US Edition; 2023-09-27 SNOMED CT UK Edition” (download shortly after the release)
“2024-02-01 SNOMED CT International Edition; 2024-03-01 SNOMED CT US Edition; 2024-04-10 SNOMED CT UK Edition” (recent download)
All dates mentioned are changed and 2 are even after the release of that vocab version. Looks like v20240229 is ANOTHER version.
Concept 37392365 is “Urine pregnancy test” and is a concept from SNOMED.
Next release 20240830 for SNOMED its the same, but in the vocab in total, is not the same. So, to me it looks like a recent version of 20240227 is an unofficial version in between the two official versions.
