A new study says human physicians are "vastly superior" to computers when it comes to diagnosing conditions and illnesses.
Is this welcome news? Sure, it's good we know one way or the other. Given the technological explosion we've witnessed and continue to experience, it's become more and more common for patients to access the internet and a growing list of smartphone apps in order to self-diagnose. And logically, such a trend has given rise to questions about the accuracy of the technology, as well as comparing those types of diagnoses to the "real" thing.
That said, is the result surprising? Not at all. Because, honestly, what other conclusion would we expect medical researchers to reach in evaluating the diagnostic performance of ... medical doctors and the medical profession?
A group of 234 internists participated in the study in JAMA Internal Medicine, titled "Comparison of Physical and Computer Diagnostic Accuracy," examining 45 clinical cases, or "vignettes." For each, the physicians were asked to list their top three possible diagnoses. And 72 percent of the time, according to the study, internists identified the illness or condition with their top choice, as compared to just 34 percent for apps and so-called digital docs. Factoring in the top three diagnoses, physicians continued to outperform technology, being correct 84 percent to 51 percent of the time.
So compared to technology, physicians performed best and made the correct diagnosis when examining less-common conditions as well as those which were more severe. Yet the performance gap between doctors and machines narrowed for illnesses that were more common, highlighting the lack of computers' analytical sophistication. The authors noted that physicians are guilty of making diagnostic mistakes -- in which they fail to identify a condition or disease, or make a correct but overly-delayed diagnosis -- in approximately 10-15 percent of all cases.
Certainly, it's comforting to learn that the results of the comparative study "show that physicians’ performance is vastly superior and that doctors make a correct diagnosis more than twice as often as 23 commonly used symptom-checker apps," according to Harvard Medical School's website. Here's hoping it will also lead to more investigations and other areas of research, and to research that doesn't involve participants examining their own profession, which would eliminate the appearance of a conflict of interest and give the public greater trust in the results.
Despite the clear existing differences between man and machine, Ateev Mehrotra, a HMS senior investigator, notes the technology has an important role in future medical work.
“While the computer programs were clearly inferior to physicians in terms of diagnostic accuracy," stated Mehrotra, who's also an associate professor of Harvard Med School's health care policy, "it will be critical to study future generations of computer programs that may be more accurate.”
