Friday, March 16, 2007

Google in medicine: Is it all that?

Medical Economics recently interviewed several experts regarding Google as a tool for medicine and the responses were mixed. Google is all the rage, but as handy as it is there are weaknesses. Google is powerful but imprecise and may be dumbing down our searching skills.

Everyone remembers the fabled BMJ paper demonstrating how Google helped lead doctors to the correct NEJM case record diagnosis in 15 of 26 cases. As interesting as that is one has to ask how it would compare with other on line differential diagnosis resources such as Isabel or Dxplain.

Dr. Eamon Armstrong, a teacher of EBM, expressed concern about the imprecision of Google and noted “’I can't think of a single person in the EBM field who would use Google on a regular basis.’”

The article drew my attention to one Google enhancement I hadn’t noted before which is the category links at the top of the results page. (This may be under development, as I found the feature for some, but not all medical searches I tried). A search of Brugada syndrome, for example, yielded this page displaying the first of 261,000 hits.




That’s too large a number to be very useful, but by using the category links at the top you can narrow the search. Clicking on “for health professionals” narrows the results down to a more manageable 386 hits.




Also note that the health professionals page provides additional categories of interest to doctors such as “CME” and “practice guidelines.”

Still, it’s imprecise. Because you never know exactly what you’re getting or how the results are chosen Google can never be better than “pretty darn good” as termed by internist and Google product manager Roni Zeiger. Because Google’s searching methods are proprietary there’s a lack of transparency, and although there are books written about numerous “hacks” you never quite know. There’s always something dark and mysterious about Google.

That’s why for more scholarly work or any search which needs to be exhaustive (e.g., has this or that type of case ever been reported in the world’s literature?) Pub Med is better. Pub Med, like Google, is powerful. But unlike Google there are no Pub Med secrets, so no hacks are necessary. It’s all out in the open and if you’re among the Pub Med uninitiated you can take their tutorial which will teach you formal search strategies. If you haven’t taken the tutorial you can still enter a simple free text search. Although that sacrifices searching power you can still know exactly what you got and how by viewing the strategy Pub Med used to retrieve your results. That’s done by clicking the details tab. Search results using the free text search syncope in patients with brugada syndrome are shown here. (I would not recommend this sloppy search statement for actual use of Pub Med but is it illustrative of how it works).

Clicking the details tab reveals what search strategy Pub Med actually used to get the results.



So which search engine is best? It depends on your objectives. Google doesn’t offer the precision of Pub Med but there are advantages, such as ability to search blogs, Wikis and emedicine. The two search engines are best viewed as complementary.

2 comments:

jon said...

Another search engine you might like to consider is the TRIP Database (www.tripdatabase.com). It is transparent, only includes high-quality material. Both PubMed and Google concentrate on quantity not quality (although PubMed has some constraints on access).

NCurse said...

Great post! I'm going to use it in my presentation on the subject (of course referring to you). Well done!