Current AI programs do worse at identifying skin lesions in people of color, research shows

Artificial intelligence (AI) in dermatology is gaining popularity for assessing, diagnosing, and treating skin conditions. However, a recent analysis of studies in the International Journal of Dermatology found shortcomings when AI is applied to skin of color (SOC). Challenges include underrepresentation of SOC in datasets, image quality issues, and lack of standardization. Current AI programs perform worse at identifying skin lesions in SOC, with only 30% of programs specifically targeting SOC. Lead author Rebecca Fliorent’s research was inspired by the lack of representation of diverse skin types in dermatology. This study highlights the need for improvement in AI programs for dermatologic purposes, especially for SOC.

Source link

error: Content is protected !!