Performance of AI, Human Readers Compared for Skin Lesion Diagnosis

WEDNESDAY, June 3, 2026 (HealthDay News) -- For skin lesion classification, a modern foundation model is better than readers with less than three years of experience, but remains inferior to experts with more than 10 years of experience, according to a study published online June 3 in JAMA Dermatology.Julien Anriot, M.D., from Claude Bernard University in Lyon, France, and colleagues compared the diagnostic accuracy of artificial intelligence (AI) algorithms versus human evaluators across varying expertise levels for skin lesion diagnosis in a multi-institutional diagnostic study. Diagnostic performance was compared among AI models and physician readers with varying dermatological expertise using a dataset of dermatological images representing everyday clinical scenarios containing 1,117 cases.Overall, 652 physicians contributed to 1,092 test iterations. The researchers found that all human readers outperformed the first-generation convolutional neural network (CNN; mean accuracy, 65.9 versus 56.7 percent). The unimodal foundation model accuracy exceeded readers with less than three years of experience (mean accuracy, 72.2 versus 68.2 percent). Experts with more than 10 years of experience achieved the highest multiclass diagnostic accuracy for skin lesion classification, with a mean accuracy of 74.2 percent -- outperforming all AI models, with accuracy of 56.7, 72.2, and 66.3 percent for CNN, the unimodal model, and the multimodal model."The future likely lies in collaboration between humans and machines to optimize diagnostic performance," the authors write.Several authors disclosed ties to the biopharmaceutical industry.Abstract/Full Text (subscription or payment may be required).Sign up for our weekly HealthDay newsletter

Performance of AI, Human Readers Compared for Skin Lesion Diagnosis

Related Stories

New, Open-Access Tool Guides Pediatric Amblyopia Diagnosis, Management

Many Seeking Skin Cancer Surveillance Care Have Low Risk for Skin Cancer

Distinct Regional Patterns Seen in Malignant Skin Cancer

Black Children Have Low Proportion of Dermatology Use