Doctors tasked with diagnosing skin diseases using images of a patient’s condition do not perform as well when the patient has dark skin. While dermatologists accurately characterized about 38 percent of images overall, they were accurate for only 34 percent of images that showed relatively darker skin. General practitioners were less accurate overall with similar decrease in accuracy with more pigmented skin.
The study, which included more than 1,000 dermatologists and general practitioners, is the first study to demonstrate physician diagnostic disparities across skin tone. Researchers compiled 364 images from dermatology textbooks and other sources, representing 46 skin diseases across many shades of skin. Most images depicted one of eight inflammatory skin diseases, including atopic dermatitis, Lyme disease, and secondary syphilis, as well as cutaneous T-cell lymphoma (CTCL).
Each of the study participants was shown 10 of the images and asked for their top three predictions for what disease each image might represent. They were also asked if they would refer the patient for a biopsy. In addition, the general practitioners were asked if they would refer the patient to a dermatologist.
Dermatologists classified 38 percent of the images correctly, compared to 19 percent for general practitioners. Both groups lost a statistically significant four percentage points in accuracy when trying to diagnose skin conditions based on images of richly pigmented skin.
Researchers gave study participants additional images to classify with an AI algorithm they had trained on about 30,000 images. Use of AI improved accuracy for both specialists and non-specialists across skin tones.
Findings appear in Nature Medicine.