Neil Pakenham-Walsh writes
> Many thanks for your message … announcing a new report on your
> excellent service Biomed News.
Thank you for your kind words. bims: Biomed News can’t fail to excel because it is a one-of-a kind service. However as such it suffers from people not understanding it. It runs against established conceptual frameworks in information retrieval. And it establishes a hitherto unknown professional practice. It takes truly exceptional individuals to pick it up.
> HIFA benefits directly from two of these 100+ reports, namely
> bims-librar on Biomedical Librarianship, and bims-skolko on Scholarly
> Communication. We often forward new papers for further discussion on
> the HIFA forums.
Happy to be useful! In addition, bims is an open data project. So it’s data is usable in bulk. This is very different from contributing say to LinkedIn or Twitter or Instagram and suchlike. While contributing to these is essentially free—you pay with your time—your contents is essentially under the control of a profit-seeking proprietary entity.
> Yes, such a report would be very useful. Do we have any HIFA members
> who would be willing to consider acting as an expert curator?
You can just start as a curator. You become an expert over time ;-)
> I cannot take on an extra commitment myself right now,
As for the time commitment, it can get down to 5 minutes a week. Obviously it will take a bit longer at the start, when the machine learning has little training data. It will permanently take more than five minutes if you sort the papers, and you regularly have more than a few. If you look at bims-skolko and bims-librar issues, you can see that I cluster the papers into what I feel are sub-areas. That takes some time. I have a third report, bims-evares, https://biomed.news/bims-evares/. Here I produce issues in five minutes, even though it is about 60 papers every week. That is because I don’t cluster papers. It’s also because the subject is full of predictable research. It’s very easy on machine learning. No need for me to even look at the top papers. I only glance at some marginal ones.
> but I would be very happy to help facilitate/support, and I’m sure
> other HIFA [members] can help.
We do have reports that are formally done by organizations or labs. I know of a duo who do it on alternate weeks.
> Another topic that comes to mind is evaluating the impact of (lack
> of) availability of relevant, reliable healthcare information,
> particularly on quality of care (across the spectrum, from lay care
> in the home and community through higher levels of the health
> system).
If you limit yourself to the topic of impact of lack of information on the quality of care, and if you exclude the aspect of education, then my hunch — and I am really out of my depth here — is that you will get few papers that are relevant. And the topic is not easy to on machine learning. But you can still get results even if you are willing to spend only five minutes a week. The approach would be to look only at very few papers, say like .07% of PubMed’s weekly input. In that case you look at about 20 papers a week. Sure with such a small number of proposed papers, you may miss out on a few papers that are relevant and that you would not be proposed to you. You could spend many weeks not selecting a single paper. Bims will improve every week by learning from the papers you did NOT select. So it will eventually proposed the right papers. This is a critical difference between bims and any system in the biomedical information sphere that came before it.
> A regular bims report looking at quality of care in LMICs, including
> a focus on the availability and application of evidence, would be a
> big step forward.
Could not agree more.
Cheers from NYC. If any HIFA members are in town, I can meet up.
-- Written by Thomas Krichel http://openlib.org/home/krichel on his 21745th day.
HIFA profile: Thomas Krichel is Founder of the Open Library Society, United States of America. Professional interests: See my homepage at http://openlib.org/home/krichel Email address: krichel AT openlib.org
