Doctors outmatch algorithms in diagnosing patients
- A new study indicates physicians are more than twice as likely as algorithms to make a correct diagnosis.
- Researchers from Harvard Medical School, Brigham & Women’s Hospital and The Human Diagnosis Project used 45 clinical vignettes to compare the diagnostic accuracy of 23 online or app-based symptom checkers with that of 234 physicians.
- The results show 72.1% of doctors listed the right diagnosis first, versus 34% of the algorithms.
Symptom checkers did better in the study published in JAMA Internal Medicine when it came to identifying the top three diagnoses, including the correct one 51.2% of the time. However, Steve Kassakian, director of clinical informatics at Oregon Health & Science University in Portland, told Medscape, “I think doctors would feel comfortable knowing that they’re not going to lose their job to a computer. But we’ve known that for decades.”
Despite beating out the machines, physicians still misdiagnosed in about 15% of the simulated cases. The vignettes were presented to internal medicine, family practice, and pediatric physicians.
A 2012 survey by Philips North America showed 49% of Americans reported feeling comfortable using online tools to understand their symptoms. And Google maintains that about 1% of all its search queries are symptom searches, which the tech giant is working to improve with help from Mayo Clinic and the Harvard Medical School.
While the findings cast doubt on the accuracy of machine-generated diagnoses, a growing number of hospitals are offering online symptom checkers. “We describe it as a right care, right time, right place tool,” Neal Linkon, director of digital engagement at Children’s Hospital of Wisconsin, told Healthcare Dive earlier this year. Yet Linkon cautioned that the tool is not meant to diagnose, but rather to help parents decide what action to take based on a set of symptoms.