Global Health Ideas

Finding global health solutions through innovation and technology

“Google Health”? Diagnosis is a keyword away

Red Herring highlighted a study published in the British Medical Journal yesterday. “How Google is changing medicine” showed that, when the two medically trained authors entered symptoms of an illness into Google, the search engine was able to correctly provide a diagnosis 58 percent of the time.

Drs. Hangwi Tang and Jennifer Hwee Kwoon Ng from the Princess Alexandra Hospital in Brisbane, Australia, wanted to see whether Google could be helpful for doctors in the diagnosis of tough cases. They were astonished to read a piece in the New England Journal of Medicine about a doctor who was able to diagnose a complicated immunodeficiency disorder by typing the information into Google.

“We think Google is likely to be a useful aid in diagnosis … It has the advantage of being easier to use and is freely available,” the doctors wrote in the study. And according to their research, Google is now used more widely than PubMed, a popular medical search engine, to find articles about medical research.


The pair selected 26 cases at random from the case records of the New England Journal of Medicine and chose between three and five symptoms that described each illness. Without looking at the conclusions of the studies, the doctors entered the symptoms as search terms into Google. The doctors then chose the diagnoses that came up most often within the first three pages of the Google search, and compared them to the actual results of each case.


In 15 out of 26 cases, the Google search provided the correct diagnosis for some pretty tough illnesses. For example, Google was able to correctly identify Creutzfeldt-Jacob disease, cat scratch disease, and lymphoma. The search engine failed, however, to diagnose pylephlebitis: It settled firmly on cirrhosis.

With search capacity moving to mobile platforms, the potential for this type of clinical second-opinion in areas with limited numbers of healthcare professionals is intriguing. A related model using menu-driven options on phones has been piloted in Kenya according to AllAfrica news (“Kenya: Health tips on your cellphone”) and blogged on NextBillion.


Written by Ben

November 11, 2006 at 10:10 pm

2 Responses

Subscribe to comments with RSS.

  1. Thanks for catching the link correction. The on-site responses that followed the BMJ article below the fold ( ”Googling for a diagnosis–use of Google as a diagnostic aid: internet based study”) get back to your point on meaning over data.


    November 16, 2006 at 1:02 pm

  2. The article’s most important contribution – having just read it (“Googling for a diagnosis–use of Google as a diagnostic aid: internet based study”) – is to highlight the tremendous power of the Google brand. The design of the study ignores the quality of the information returned. The authors match search results with symptoms, but do not attempt to distinguish among the sources of information. That is left to the machines. In this experiment, they are treating Google as a diagnostician, rather than as diagnostic support. It’s not really appropriate to do that now, but once Web 3.0 comes around it may be. Web 3.0, or the semantic web, as Nicholas Carr states, will be about “mining ‘meaning,’ rather than just data.” Describing how that will happen is perhaps a bit much to explain here, but there are plenty of explanations out there on the web.


    November 16, 2006 at 5:04 am

Comments are closed.

%d bloggers like this: