When I was a graduate student at Columbia University, I had the extraordinary privilege of working in the same building as one of the world's greatest medical research libraries. When I saw a reference to a paper I wanted to read, I could usually lay my hands on a copy of it within minutes, no matter how obscure the journal or how long ago it was published. Often, the slowest part of the process was waiting for the elevator.
Besides hosting miles of shelves packed with the combined output of hundreds of medical journals, the library also had a lounge with the latest issues of all of those same journals. Whenever I was sufficiently motivated (which was less often than it should have been, but still), I could sit in a comfy chair and catch up on Science, Nature, or the Journal of Virology.
Today, of course, we have the Internet. It's not always better. Yes, the world of online ressearch is faster and vastly more accessible than the basement of the Hammer building, assuming one has keys for the Web's various walled gardens. That's fine for very focused reading. If I want to find all of the papers about Zika virus and microcephaly published in the past year, I can look them up in milliseconds, and start digging through them moments later. That's lovely.
What's missing, though, is the serendipity. Whenever I looked up an old paper deep in the stacks, I would pull a volume off the shelf, blow away some dust, and open the pages to the original publication, bound in its journal just as it was when it came out. It was quite common to glimpse the paper right before or after it, realize that the publisher had run them together in the pages for a reason, and decide that this, too, was worth reading. The same thing happens when paging through the latest issue of any printed journal.
The Internet excels at answering the questions we ask it. Where it fails is in telling us what questions we ought to be asking.
Individual journals have tried to combat this with helpful suggestions alongside articles, but the implementation varies tremendously, the interfaces are often clunky, and in many cases they're just basing what "You might also like..." on an algorithm no more sophisticated than the PubMed search that brought the reader there in the first place. There is, however, another way.
It's called Medical Subject Headings (MeSH). Readers who are already deeply enMeSHed may not get much from this post, but on the assumption that I'm not the last person to figure out how useful this system is, I'll continue.
When looking at the abstract of an article on PubMed, see if one of the expandable options below it mentions "MeSH terms." If so, expand it, click one of the terms, and choose the option to search PubMed for that. The result will be a human curated bibliography of related papers. Yes, that's right, actual people, specially trained to read, understand, and index the biomedical literature, studied all of those papers and placed them under that heading just for you to find. They're still doing it too, so if you bookmark that MeSH search and come back to it later, the bibliography will be updated.
It's worth reading up a bit on how this whole process works, to understand its strengths and limitations. There are also some useful related tools like the MeSH Browser, which provides a browsable tree of headings and subheadings. Clicking through the MeSH Browser is the digital equivalent of wandering the stacks. If you'd like to know where your own work fits into this hierarchy, but none of your publications have been indexed yet, try pasting some text into the MeSH on Demand tool.
Because it's done by people, there's a lag in MeSH indexing. It's not a uniform lag, either. Publications in journals with high impact factors tend to get indexed faster than lower-impact publications, for example, and there are probably differences between fields as well. If you need up-to-the-minute updates on what your competitors are doing, MeSH monitoring isn't the right tool. For keeping up or catching up with a broad section of the literature, however, it's perfect.
It also combines well with the other options on PubMed. For example, I cobbled together a list of my favorite MeSH headings into a search string, then clicked "Create RSS" under the search box. That gave me a human-generated news feed of primary research publications spanning a huge swath of the literature. I'll talk more about RSS in another post, but if you're already using a service such as Feedly, you know how powerful news feeds can be.
I still miss the magic of flipping a page and finding something wonderful on the other side. My MeSH searches bring no low thrumming of basement machinery juxtaposed with the mechanical sizzle of the light timer at the end of the aisle, no muffled sound of distant voices, no long canyons walled in by half-forgotten tomes. But there's also no closing time, no malfunctioning photocopier, and no key card. The stacks may be long gone, but the MeSH is always open.