Deep Learning Improves Diagnostic Accuracy for Skin Diseases

FRIDAY, Feb. 16, 2024 (HealthDay News) -- Deep learning-aided decision support improves diagnostic accuracy for skin disease, according to a study published online Feb. 5 in Nature Medicine.

Matthew Groh, Ph.D., from the Northwestern University Kellogg School of Management in Evanston, Illinois, and colleagues presented results from a study involving 389 board-certified dermatologists and 459 primary care physicians, which assessed the accuracy of diagnoses submitted by physicians in a store-and-forward teledermatology simulation. Physicians were asked to submit up to four differential diagnoses for 364 images spanning 46 skin diseases.

The researchers found that the diagnostic accuracies were 38 and 19 percent, respectively, for specialists and generalists. For diagnosis of images of dark versus light skin, both specialists and generalists were 4 percentage points less accurate. The diagnostic accuracy of both specialists and generalists was improved by more than 33 percent with fair deep learning system decision support; the gap in diagnostic accuracy of generalists was exacerbated across skin tones.

"This study allows us to see not only how artificial intelligence assistance influences, but how it influences across levels of expertise," Groh said in a statement. "What might be going on there is that the primary care physicians don't have as much experience, so they don't know if they should rule a disease out or not because they aren't as deep into the details of how different skin diseases might look on different shades of skin."

One author disclosed ties to the biotechnology industry.

Abstract/Full Text