An article published in BMJ argues that Google searches can sometimes aid with developing an appropriate diagnosis of disease.
- Hangwi Tang, Jennifer Hwee Kwoon Ng. Googling for a diagnosis—use of Google as a diagnostic aid: internet based study. BMJ 2006: 333; 1143-1145. Available in html format or pdf format.
I posted a summary of this article in Chance News 23.
The authors discussed the need for highly specific search terms and used the wording “statistically improbable phrase” which was a concept popularized by the Amazon web site. I wrote a brief description of the concept of statistically improbable phrases, also in the same issue of Chance News.
I didn’t want to brag about my book on Chance News, but here I am less inhibited. The SIPs for my book, Statistical Evidence, are
- full free text,
- total respiratory heat loss,
- open source publication,
- voice handicap index,
- clinical indifference,
- covariate imbalance,
- alternating assignment,
- spectrum bias,
- funnel plot,
- diagnostic distribution,
- suicide completers,
- noncompliant patients,
- breastfeeding practices,
- teenage smokers,
- multiple outcome measures,
- cocaine abstinence,
- publication bias, and
- body fatness
I’m not sure that all of these are helpful, but for an automated algorithm with no human intervention, the choices are not bad. It would be interesting to look at SIPs for some classic statistics books as well as some classic works of fiction.