Keeping track of publications
We track physician publications as part of the CV and performance / research review. Rather than doing this manually and repeatedly in different formats, it would be nice to do it once and re-use the information.
- physicians and/or their secretaries keep track of these manually
- sounds like some actually tell the secretaries and some secretaries monitor Pubmed
Attempt at automation
- Template:PersonAutolister attempts to provide a link to a pubmed search by the name for the person on the wiki; this works reasonably well for rare names, but not so much for rarer ones.
We use SciVal to aggregate metrics for our physicians. It appears this tool bases its data on ORCID IDs and Scopus IDs.
Scival is available only from the U of M network, and it requires a login. Tom Fraser is our resource for this.
Tom spends time at annual report time to combine records for physicians who publish under multiple names. While this can be done in SciVal, the better place would be to combine them at their source.
U of M Libraries efforts
U of M Libraries is also working on this. Tom and Tina met with Janice Linton to discuss. Turns out that libraries has gained some admin level access to Scopus to facilitate merging of authors there. Also, there is an institutional membership in OrcID. The lead on this is Andrea Szwajcer, and we have requested a meeting with her. User:Ttenbergen 16:08, 2019 August 13 (CDT)
- Is there a way to uniquely identify physicians? Can either be used to identify them on pubmed?
- Several systems attempt to do this, but for now, none are used consistently enough to work.
- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5451540/ in 2016 describes that there is no simple solution to this even though it's a big deal.
Pubmed searche results can be output as a RSS feed. With Extension RSS Mediawiki can render incoming RSS feeds (on a totally unrelated note, this would allow us to integrate our blog on the wiki...). This would provide a listing, but no data we can count or otherwise manipulate.
Extensions for Pubmed
We are not the first with this requirement. The following extensions are available to wrangle Pubmed data into mediawiki:
- https://www.mediawiki.org/wiki/Extension:Pubmed - provides a list of publications
- https://www.mediawiki.org/wiki/Extension:PubmedParser - provides structured data about publications
Via Extension:Cargo and Extension:External Data
- https://www.mediawiki.org/wiki/Extension:External_Data - can link into external data e.g. in JSON format
- https://www.ncbi.nlm.nih.gov/pmc/tools/get-metadata/ - provides data e.g. in JSON or XML format
How could we do this?
- We won't be able to automate this completely because we can't identify authors reliably
- So, we will need to continue to review Pubmed for publications and get the PMIDs from there
- Once we have the PMID we could list those manually on a physician's page.
- Question: Would we be able to import the bulk of the page's content from there, and treat it as data internally that we could count or similar?
Should we store this on our wiki or is there a better place?
If we are going to go through the trouble of managing the publications, can we do it somewhere more powerful than on our piddly little wiki?
Turns out Wikidata already stores much of this! If we are going to curate it anywhere, we should curate it there and make the world a better place. We would need to generate author entries for our physicians (the wikidata people might be able to help us with an import) and publication entries where they don't already exist.
- Pro: Would fix the problem at the root.
- Con: is even nastier to edit than our wiki. :-(
What format do we actually need this in?
Apparently we use this in several places
- personal resumes
- Dept Head updates
What formats do we need for all of this, to make sure we have the relevant data?
Data fields required
Internally, we need to know the following about publications:
- published in
- PubMed or DOI (which is preferred?)
- is primary author
What do librarians think?
I talked with some librarians at U of M and they said that the U of M is looking into getting a system set up to manage academic research and publication information better. This sounds like it would be several years out at best. An open-source example of the types of system they would be looking at is [VIVO, which can facilitate the harvesting of data from Pubmed etc.