Morayo Adisa, MD, is a double board-certified dermatologist, dermatopathologist, and president of the Chicago Dermatological Society. She is also the medical director of Dermatology Physicians SC in Chicago and Kenilworth, Illinois.

Morayo Adisa, MD

Skin of Color Savvy: The Art and Science of Treating Patients of Color

Skin of Color Savvy: Tech Check—AI, Telederm, and the Bias Question

Skin of Color Savvy: The Art and Science of Treating

, Morayo Adisa, MD, medical director of Dermatology Physicians SC in Chicago and Kenilworth, Illinois, sat down with Roxana Daneshjou, MD, PhD, assistant professor of biomedical data science and dermatology at Stanford University, to examine one of the most rapidly evolving topics in medicine: the intersection of artificial intelligence, 

, and equitable care for patients with skin of color.

AI systems can perpetuate existing healthcare disparities if trained on biased or nonrepresentative datasets.

Improving dermatology education and diversifying AI development are critical to achieving equitable care.

Teledermatology expands access but must address challenges including camera bias, digital access, and patient trust.

, featured a discussion exploring both the promise and limitations of artificial intelligence (AI) in dermatology, with Daneshjou emphasizing that while AI has the potential to improve healthcare delivery, it also risks perpetuating longstanding disparities if not developed and evaluated thoughtfully.

Daneshjou explained that many AI systems inherit biases already present within healthcare because they are trained on historical data that reflect existing inequities. She highlighted research demonstrating that algorithms have, in some cases, allocated healthcare resources inequitably by relying on spending data rather than true disease burden, inadvertently disadvantaging Black patients. Within dermatology specifically, she discussed her own work evaluating skin cancer detection algorithms, which found significantly poorer performance when identifying malignancies on darker skin tones compared with lighter skin.

A central theme of the episode was that AI cannot solve problems rooted in deficiencies within the healthcare system itself. Daneshjou noted that dermatology education has historically underrepresented skin of color in textbooks, educational resources, and training materials. As a result, clinicians may diagnose conditions later in patients with darker skin, limiting the availability of high-quality, early-stage clinical images needed to train more equitable AI systems. Improving AI performance, she argued, must begin with improving dermatology education and increasing representation across all aspects of clinical training.

The conversation also examined practical steps dermatologists can take to improve the future of AI. Daneshjou advocated for diversifying image datasets, ensuring broader representation among researchers and technology developers, and routinely evaluating algorithms for fairness before widespread implementation. She stressed that equity should be considered during the development process rather than only after products reach the market.

Adisa and Daneshjou also discussed the expanding role of teledermatology, acknowledging that virtual care has increased access for many patients while introducing new challenges. Although teledermatology can improve access for individuals with transportation or mobility limitations, Daneshjou cautioned that disparities persist through unequal internet access, language barriers, and differences in smartphone camera technology. She noted that camera systems have historically been calibrated using lighter skin tones, potentially reducing image quality and diagnostic accuracy for patients with darker skin during store-and-forward teledermatology encounters.

Patient trust emerged as another important focus. Daneshjou emphasized that patients should be informed whenever AI is incorporated into their care, whether through documentation tools, clinical decision support, or other technologies. She advocated for explicit patient consent surrounding AI use, particularly given evidence that current systems may not perform equally across diverse populations.

The discussion also touched on AI scribes, which have become increasingly common in clinical practice. Although available within her own institution, Daneshjou explained why she has chosen not to adopt the technology, citing concerns about patient privacy, inaccurate documentation, hallucinated clinical information, and the importance of personally crafting assessment and treatment plans as part of her own clinical reasoning process.

Looking toward the future, Daneshjou acknowledged that AI will almost certainly become more deeply integrated into healthcare but stressed that ongoing advocacy, rigorous evaluation, and greater diversity among developers will be essential to ensuring these technologies improve rather than worsen health disparities. She also identified textured hair and hair disorders as an area where AI research remains notably sparse, presenting an opportunity for future innovation in skin of color dermatology.

The episode concluded with a call for continued collaboration among clinicians, researchers, technology developers, and organizations such as the Skin of Color Society to ensure that advances in AI and digital health are guided by principles of equity, transparency, and representation, ultimately benefiting all patients regardless of skin tone.

To learn more about SOCS’s programs and initiatives, visit 

Editor’s note: This episode was summarized with the help of AI tools.

Kaundinya T, Kundu RV. Diversity of Skin Images in Medical Texts: Recommendations for Student Advocacy in Medical Education. J Med Educ Curric Dev. 2021 Jun 11;8:23821205211025855. doi: 10.1177/23821205211025855. PMID: 34179498; PMCID: PMC8202324.

Podcasts

In this podcast episode, 2 featured experts discuss AI, teledermatology, and strategies to reduce bias in skin of color dermatologic care.

Morayo Adisa, MD

Articles