Diagnosing skin conditions in patients with skin of colour: Is colour the only challenge?

Abstract

Inflammatory skin diseases can manifest differently across diverse skin types. Clinical images and data representing individuals with skin colour (SoC) remain limited, reflecting their underrepresentation in medical education and dermatological research.

In this issue of JEADV, Shencoru et al.¹ surveyed 132 medical students using images of common dermatoses across different skin types. Diagnostic accuracy was lower in SoC, particularly for urticaria and seborrheic dermatitis (SD), raising important questions about whether SoC constitutes a diagnostic challenge and how dermatology training should balance core principles—such as morphology, distribution and pattern—with variations in clinical presentation across skin tones. Students were more accurate in lighter skin, suggesting that reduced visibility of erythema in SoC may hinder recognition. While erythema is more apparent in lighter skin, urticarial lesions in SoC may appear skin-coloured, darker or greyish-purple. Nevertheless, the defining feature is the presence of wheals, which should be recognized regardless of skin tone.

SD typically presents with erythema and scale; however, in darker skin, it may also appear as hypopigmented patches with subtle erythema. Arcuate or petaloid lesions along the hairline (‘petaloid seborrheic dermatitis’) may occur, often with minimal scaling. Other inflammatory skin conditions, such as atopic dermatitis, may present as follicular eczema or with post-inflammatory dyspigmentation in individuals with SoC. Familiarity with these variable presentations is essential for accurate diagnosis.

Emerging artificial intelligence (AI) technologies may support dermatology education, but important limitations remain. Recently, Joerg et al.² evaluated four leading text-to-image generators and found poor clinical accuracy alongside limited representation of diverse skin tones, reflecting biases in training data. The study also highlighted the limitations of relying solely on the Fitzpatrick skin phototype scale and emphasized the need for more representative datasets. Alternative classification systems, including the Monk Skin Tone Scale, the Individual Typology Angle (ITA) and the Eumelanin Human Skin Colour Scale, have been proposed to better capture skin tone variation.³

To improve clinical care, dermatology training must be more inclusive, combining greater exposure to SoC with a structured emphasis on lesion morphology, distribution and pattern recognition. International initiatives, including those from the International Psoriasis Council (IPC) and the International League of Dermatologic Societies (ILDS), have developed image galleries and atlases reflecting diverse populations. Additional resources, such as the JAAD SoC Image Atlas and recent textbooks—including Taylor & Kelly's Dermatology for Skin of Color (2026) and Taylor & Elbuluk's Atlas and Synopsis for Skin of Color (2023)—further enhance representation in teaching materials. Undergraduate curricula in countries such as the United States and United Kingdom increasingly incorporate SoC-focussed content, complemented by supplementary educational programmes offered by professional societies, including the Skin of Color Society and Skin of Color Training UK.

Importantly, SoC education should also address cultural and socioeconomic factors that influence access to care. Patients with darker skin often experience delayed diagnosis, more severe disease and a greater impact on quality of life due to systemic healthcare biases and barriers to treatment.⁴ Another persistent challenge is the lack of a standardized definition of SoC, which is used as an “umbrella” term and may limit consistency in research and education. Definitions should consider genetic, ethnic and other relevant factors. Recent international consensus efforts, such as those proposed by Lim et al.,⁵ aim to provide a clearer framework to support equitable dermatologic care.

Overall, these findings underscore the need to adapt dermatology education and practice to better reflect diverse populations. Emphasizing on morphology, distribution and secondary changes—alongside improved representation in educational materials—can enhance diagnostic accuracy and reduce disparities. Generative AI may serve as a valuable adjunct, but its effectiveness will depend on addressing current limitations in data diversity. Combined with broader systemic efforts, these strategies are essential to advancing equity in dermatologic care.

None.

None for the authors to disclose.

Not applicable.

Not applicable.

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.