Decide how to handle missing/deprecated PMIDs

[andrewsu]

As raised by @bill-baumgartner in https://github.com/NCATSTranslator/Feedback/issues/625#issuecomment-2237699515 and https://github.com/NCATSTranslator/Feedback/issues/803#issuecomment-2225869209, there are cases where KPs are returning PMIDs that are no longer in PubMed. Creating this issue to track how this should be handled.

[andrewsu]

@Genomewide asked:

How do we get the KPs that provide PMIDs to filter this? We can not do anything about this? @andrewsu Do you filter these out? Are there other teams that need to look at this group to filter that you know of?

I think there are two questions -- how to solve this problem, and what's the priority of this issue.

• Potential solutions: I think this should be some sort of centralized service. PMIDs are used everywhere, and delegating that to every KP to do themselves will undoubtedly lead to inconsistency.
• Priority: I think 120k examples of missing/deprecated PMIDs out of 37 million total in PubMed is a pretty low incidence rate. So unless there is some reason that we hit these a lot, I think this is a low priority item. (Note, retracted articles do not have the PMIDs removed, so this is due to something else. But a microservice to check for deprecated PMIDs might also check for retracted articles to flag...)

[Genomewide]

That sounds reasonable to me. I was not thinking about the small fraction that it is. Agree that this is likely a low priority. This is the first time I have seen a ticket for it. Microservice would be great. Do you want to call this closed or snoozed?

[sierra-moxon]

I'm going to "next phase" it.

[codewarrior2000]

@andrewsu Earlier this year in February, I wrote a little code to check whether PMIDs were retracted or had errata. Extending that code to check for missing PMIDs and then wrap it all up into a microservice sounds intriguing.