Is your clinical job safe from automation?
Well, the good news is that doctors beat the robots more than two to one in a recent study.
When it comes to work, humans have always competed with machines.
Email took over for snail mail. Automatic toll collection replaced highway toll takers. Cars can now drive themselves, (sort of). So, there’s certainly reason to think that machines could one day overtake humans when it comes to diagnosing disease. Right?
Researchers from Harvard Medical School and Brigham and Women’s Hospital in Boston decided to put humans and machines to the test in the first of its kind head-to-head study published in the Journal of the American Medical Association. The study compared the diagnostic accuracy of more than 230 physicians with that of 23 different medical diagnostic software tools—including WebMD, the Mayo Clinic’s Symptom Checker, AskMD and Drugs.com. Both the physicians and the automated systems were fed medical scenarios and were asked to make and rank three possible diagnoses for each.
These 45 scenarios, or “medical vignettes” as they were referred to in the study, ranged from relatively simple to much more complicated maladies—everything from canker sores to appendicitis. In each scenario, participating humans and machines were provided a description of the symptoms and the patient’s medical history but were not told any results from physical examinations or medical tests. Twenty physicians diagnosed every vignette.
The results: 84% of the doctors named the correct diagnoses in the top three possibilities, compared to computers, which only hit the mark 51% of the time. On their first tries, 72% of the doctors pinpointed the correct diagnosis, compared with 34% of the digital platforms.
However, doctors noted a couple of factors that could impact the results of the study.
For one, Ateev Mehrotra, an associate professor of health care policy at Harvard Medical School and the senior study investigator, told CBS News that he believes the degree of difficulty had an impact on the digital platforms’ accuracy, as the results showed that the online symptom checkers were more likely to make a correct diagnosis for simpler health conditions than more complex ones.
And as for the physicians in the study, Keri Peterson, an internist at Lenox Hill Hospital in New York City, told CBS News that had the physicians been able to see the patients, there would have most likely been even fewer misdiagnoses.
“I think the physicians’ accuracy in the study would go up even higher if they had access to the patient,” Peterson said. “We get so much by just walking into the room and seeing them and looking at how sick they look. I don’t see that computers would ever replace doctors.”
So fortunately for you health care workers, it looks like you’ve got the robots beat. Though, Mehrotra said that’s not to say one day computers couldn’t beat out clinical workers…just that day won’t be any time soon at least.