Artificial Intelligence (AI) Tools for Dermatology Face Challenges in Skin of Color
A recent study published in the International Journal of Dermatology highlights the challenges faced by artificial intelligence (AI) tools in dermatology, particularly in their application to skin of color (SOC). Researchers from the Rowan-Virtua School of Osteopathic Medicine in New Jersey found that one of the main obstacles is the underrepresentation of SOC in datasets.
Analyzing literature from the past decade, the study identified issues with the quality and standardization of images, which hinder the efficacy of current AI programs in identifying skin lesions in SOC. Only 30 percent of the programs reviewed in the study specifically addressed their use in dermatology for SOC patients.
The authors note that while many AI programs show promise in supporting clinical decision-making, significant development is needed to ensure their reliability as diagnostic aids for SOC patients. Improvements in the representation of darker skin tones in datasets are crucial for the accurate detection and treatment of skin conditions in individuals with SOC.
As the field of AI in dermatology continues to evolve, addressing these challenges will be essential for advancing the use of technology in providing effective and equitable care for all patients. Further research and development are needed to enhance the capabilities of AI tools for dermatological applications, especially in addressing the unique needs of SOC individuals.
