Standard Device Vocabularies

[PRijnbeek]

I see there are now many vocabularies that have standard codes in the device domain:

However, not all of these are selected by default in ATHENA: Why is that?

What is the practice now in this domain, if you cannot find a standard you make your own local source codes the standard? and then all the future vocabularies have to try to map to yours? So if we cannot map to standard device codes in IPCI (if we would be aware of all the vocabs that are standard), we simple add those as a new vocabulary and call them standard?

I see for example the Japanese JMDC has standard codes, in the device domain which look like drugs by the way:

Why are these not added in RxNorm Extension if they are missing?

[cgreich]

However, not all of these are selected by default in ATHENA: Why is that?

If none of the vocabs are selected, all are selected. The moment you click on one the others are no longer included. I agree, that is confusing, but it is actually very efficient once you figured it out.

What is the practice now in this domain, if you cannot find a standard you make your own local source codes the standard? and then all the future vocabularies have to try to map to yours?

For devices, we have no standard system, yet. No hierarchy for that matter, either. All of them, if not deprecated by the source, are standard. We are talking to SNOMED about that problem.

So if we cannot map to standard device codes in IPCI (if we would be aware of all the vocabs that are standard), we simple add those as a new vocabulary and call them standard?

No mapping. You make your IPCI devices standard. We will later figure out how to map them to some standard.

I see for example the Japanese JMDC has standard codes, in the device domain which look like drugs by the way. Why are these not added in RxNorm Extension if they are missing?

Barium sulfate is not a drug. It is contrast material. Drugs have to have a biochemical effect on the metabolism. Radiopharmaceuticals same thing, btw, they are also devices. So are tissues and blood products. There is debate to change that in the Forum.

[PRijnbeek]

Thanks helpful to learn more about this domain.

If none of the vocabs are selected, all are selected. The moment you click on one the others are no longer included. I agree, that is confusing, but it is actually very efficient once you figured it out.

Not sure I fully understand this answer. If you click on download in Athena some are selected, what is the logic these are selected by default? I would expect that all the vocabularies are selected that have standard codes but maybe there is another logic I am not aware of?

No mapping. You make your IPCI devices standard. We will later figure out how to map them to some standard.

I assume you should still try to map to those that are already standard first right so you need all of standard device vocals in Usagi? The challenge is that users do not know which are all those vocabularies without searching since they are not selected in the Athena download.

[Alexdavv]

I would expect that all the vocabularies are selected that have standard codes but maybe there is another logic I am not aware of?

Your logic is right, but there are some exceptions. Devices (the whole Domain) is one of them. Another one is EAV-type data where we didn't figure out yet what to do. As the result in the Drug Domain, only proper target vocabularies are selected by default + NDC/SPL (due to its high utilization, I think). If you don't find in your source any specific drug vocabularies (dm+d, Gemscript, DPD) or UKBiobank, PPI, etc, you won't map to them, even though some standard concepts exist. They are standard because we don't have a better solution yet. Not because they're good targets.

I assume you should still try to map to those that are already standard first right so you need all of standard device vocals in Usagi? The challenge is that users do not know which are all those vocabularies without searching since they are not selected in the Athena download.

I don't think there's any benefit. You'll get an ugly mixture of 5-10 vocabularies with no logic/hierarchy between the concepts. The only thing that makes more sense is to map to the source vocabulary if it was used in the data, but the codes are missing and you try to recover them using the descriptions.

[alepev]

Hello, jumping in the conversation. Could it be helpful to provide Athena users with a list of recommended standard vocabularies, and "lack of better choice" standard vocabularies to guide mapping choices? Perhaps a link at the top of the Download page in Athena? (you could keep a curated .md file here on GitHub and just link to it)