CEO of VisualDx touches on a decades-old blind spot in diagnosing skin of color | Fall Clinical Derm 2026

Most medical image libraries underrepresent darker skin tones. That gap can make it harder for clinicians to spot key signs of disease, according to Art Papier, M.D., CEO of VisualDx.

VisualDx, which makes a clinical decision support tool, has collected images of people of all skin tones since it started in 1999, Papier said.

Common signs of illness can look very different depending on skin tone, he said. Increased blood flow shows up as redness on light skin, but on brown skin it looks like a deeper brown. Purpura, a sign of blood leaking out of blood vessels, is often linked to serious illnesses. It appears purple on white skin but very dark on people of color.

Papier said even some dermatologists who trained in areas with few African American patients and people of color struggle to recognize these signs. He said the need for these tools is greatest in primary care and emergency departments.

The gap in training is not new. In 2006, Papier and colleagues published a paper in the Journal of the American Academy of Dermatology on disparities in dermatology resources. After reviewing textbooks, resources and lectures, they found almost all teaching featured light skin.

Papier said the company’s focus on skin tone did not start with the renewed attention to health equity after the killing of George Floyd in 2020. It grew out of a broader question about why clinicians make mistakes.

Skin tone bias is one of several cognitive biases that software developers have to consider, he said. Gender bias is another. In internal medicine, assumptions that women don’t have heart attacks have led to those heart attacks being missed more often.

“So what we do is we look at: How do clinicians problem-solve? What are the most frequent mistakes? And then how do we fix those mistakes? And so we work backwards from the known errors,” Papier said.

Diagnosing conditions in skin of color has been a known problem since he started in dermatology 30 years ago, Papier said. VisualDx continues to work on it along with other biases in clinical decision-making.