Google Searches Out the Diagnoses, Stat

BRISBANE, Australia, Nov. 10 -- Google can pluck diagnoses for tough cases out of the range of stumped into the realm of inspired in minutes, with a decent grade for accuracy, found researchers here.

The world's most popular search engine returned a correct diagnosis 58% of the time, said Hangwi Tang, M.D., and Jennifer Hwee Kwoon Ng, both of the Princess Alexandra Hospital here, in a study reported online in BMJ.

The researchers fed Google three to five search words from the prior year's New England Journal of Medicine diagnostic case reports. They got back links suggesting a correct diagnosis for 15 of the 26 cases (95% confidence interval 38% to 77% correct).

"Useful information on even the rarest medical syndromes can now be found and digested within a matter of minutes," they concluded. "In difficult diagnostic cases, it is often useful to 'Google' for a diagnosis."

While most effective for conditions with unique symptoms and signs that can easily be used as search terms, the authors said Google may not be as prescient at diagnosing complex diseases with non-specific symptoms or common diseases with rare presentations.

They cited a case of ehrlichiosis for which Google suggested amyotrophy, vasculitis diagnosed by Google as uveitis, and a rare presentation of endometriosis that Google suggested was tuberous sclerosis.

Before reading the NEJM's diagnosis, the researchers examined the top 30 to 50 search results from Google using search terms selected from the article. They selected the diagnosis that best fit the case record or up to the top three most likely diagnoses when none fit well. If one was correct according to the NEJM's diagnosis, the search was deemed successful.

Diagnosis is challenging and fulfilling, the investigators said, but relies on skills search engines are best at, reigning in a massive knowledge base.

"Physicians have been estimated to carry two million facts in their heads to fulfill this role," they wrote. "With medical knowledge expanding rapidly, even this may not be enough. Google gives users ready access to more than three billion articles on the Web."

While many Internet-savvy doctors have discovered the search engine's utility in diagnosing tough cases, so have patients. Dr. Tang said a patient's father sparked her interest in the accuracy of Google diagnosis searching.

"After evaluating a 16-year-old water polo player who presented with acute subclavian vein thrombosis, one of us started to explain that the cause of the thrombosis was uncertain," she said.

"But of course he has Paget-von Schrtter syndrome," the father blurted out.

"Having previously googled the symptoms," she said, "he gave us a mini-tutorial on the pathophysiology (hypertrophy of the neck muscles leading to dynamic compression of the axillary vein at the thoracic inlet-leading to thrombosis) and the correct treatment of the syndrome."

However, search engines are nowhere close to pay for performance.

"Arguably, everything could be found on the Web if only one knew the correct search terms," but choosing the correct search terms and the correct information out of the "noise" can require an expert, the investigators found.

"The efficiency of the search and the usefulness of the retrieved information also depend on the searchers' knowledge base," Dr. Tang and Hwee Kwoon Ng said. "Patients doing a Google search may find the search less efficient and be less likely to reach the correct diagnosis."

They suggested that "Google searches by a 'human expert' (a doctor) have a better yield, as Google is exceedingly good at finding documents with co-occurrence of the signs or symptoms used as search terms and human experts are efficient in selecting relevant documents."

In evaluating the search results for the study, the researchers said they attempted to avoid using specialist knowledge, instead choosing diagnoses that were ranked most prominently and seemed to fit the case record.

In some cases, Google returned a diagnosis that was correct but not sufficiently specific. For example, a case of "hot tub lung" (extrinsic allergic alveolitis caused by Mycobacterium avium) was diagnosed as extrinsic allergic alveolitis by Google, which the authors called an incorrect result.

They said that "Googling" for a diagnosis is likely to become even more useful with the further development of Google Scholar, which searches only peer-reviewed articles.

"Web-based search engines such as Google are becoming the latest tools in clinical medicine, and doctors in training need to become proficient in their use."