A study suggests that doctors are way better than symptom-checker points at reaching a correct diagnosis. A head-to-head comparison showed that human doctors with access to the same info about medical history and symptoms as a symptom checker had a 72 percent success in reaching the right diagnosis. For apps, it was just 34 percent.

23 online symptom checkers were scrutinized in the study. Out of these, some could be accessed via websites and others were available as apps. These included leading symptom checkers like those offered by Web MD and the Mayo Clinic.

The researchers used a platform to distribute 45 clinical vignettes to 234 physicians. The doctors who couldn’t do a physical examination had to rely entirely on the information provided to make a diagnosis. Fifteen of the vignettes described acute conditions while 15 were only moderately serious. Another 15 called for low levels of care. Most of the vignettes described commonly diagnosed conditions. 19 vignettes described uncommon conditions. The doctors submitted answers as free text responses. They ranked potential diagnoses in order of likelihood.

When compared to putting the same information into symptom checkers, the doctors ranked the correct diagnosis first more on every case. Also, the doctors got it right more times for the more serious conditions and also for the uncommon diagnoses. But it was found that the computer algorithms were better at finding out less serious conditions and also common diagnoses. The results were published in a research letter in JAMA Internal Medicine.

The study also suggests that since humans are prone to make errors, they could benefit from using the algorithms to augment their skills.

Image credits: npr.org

   Send article as PDF