NEJM Image Challenge and (Lack of) Wisdom of Crowds



The New England Journal of Medicine Image Challenge "provides an opportunity for you to use the clinical images from NEJM to test your diagnostic skills.

A new image and question appears each week, along with several possible answers. Before selecting an answer, you can take a closer look by zooming in and moving the image within the frame. You can also look at the relative distribution of answers chosen by other users." Source: NEJM.

The Image Challenge is a good review tool for USMLE, ABIM or residency in-service exams. When I studied for USMLE, I used to group clinical images from NEJM under system headings, e.g. Cardiology, Pulmonology, etc. but the image quiz/challenge is more fun.

The NEJM Image Challenge is, in a sense, a Web 2.0 tool -- it relies on user-created content in terms of recording readers' responses and comparing them to correct answers. It is surprising to see that for many image challenges the readers ("the crowd") did not get the right answer despite the massive participation of tens of thousands of responders for some queries -- click to enlarge the image above for an example. This may be an indication that for professional fields like medicine the "wisdom of crowds" concept does not work as well as, for example, Digg.com or Wikipedia. I would prefer an expert's opinion over the crowd's opinion most of the time, I think.

It could be an interesting mini-research project to have a look at all the answers in the NEJM Image Challenge and to see how many times the "crowd" got it wrong versus the "expert" (NEJM) who always provided the correct answer. In August 2007, I calculated the percentage of correct answers and the number of responses per question from all the images available on the NEJM website at the time. The results showed that there were on average 8300 responses per question and the "crowd" solved correctly only 69% of the image challenges.

Related reading

JAMA liked NEJM's Image Challenge so much that they launched their own "Clinical Challenge" http://jama.ama-assn.org/site/clinicalchallenge/jcg-index.xhtml
Make the Largest Encyclopedia in the World Better. I already did (I hope), 9/28/2005.
You know NEJM has a clinical "Image Challenge" right? Now The Lancet has "Picture Quiz" http://goo.gl/1NLIq - Where are you, BMJ?
Digg for Medicine: Nature Publishing Group Launches Dissect Medicine, 5/03/2006.
Edit Wars in Medical Wikipedia, 12/09/2006.
Google Finds Correct Diagnosis in 58 % of Cases Published in NEJM, 11/11/2006.
The only way to preserve the wisdom of the crowd is to protect the independence of the individual. WSJ, 2011.
NEJM Image Challenge is one of the most popular features on NEJM.org, and now there is a new way to play - on Facebook http://goo.gl/3VfGf

2 comments:

  1. For true wisdom of the crowds to be in effect, it is important to prevent an information cascade. Allowing the user to see the votes of others before they vote leads to biased results.

    It is also not clear how diverse the crowd is in this cases, as anyone can attempt the answers. The crowd needs to have some insight, so non-medical users may also bias the result.

    ReplyDelete
  2. I have to second what anonymous said - I do some of these image challenges from time to time just as a fun way to introduce myself to medical concepts and conditions I'm unfamiliar with. As a medical librarian rather than a physician, I'm quite sure I'm skewing the results to the incorrect side (although I do get quite a few correct!). :)

    ReplyDelete

Blog Widget by LinkWithin