People with skin of color face a clear gap in acne AI: most tools were built and tested with little dark-skin data. You can compare acne apps by checking skin-tone testing, expert review, and regulatory status before trusting a result.
Acne AI means phone apps and online tools that grade acne from photos. A 2025 scoping review of 105 acne-AI papers found only 13% reported skin color or ethnicity, per the International Journal of Dermatology 2025 review of 105 acne-AI papers. Only 3.8% included skin-tone data when building models.
Medical information disclaimer: This article is for general educational purposes only and does not provide medical advice, diagnosis, or treatment. Always consult a physician or other qualified health professional about symptoms, medications, tests, or treatment decisions.
Table of Contents
- Why does accuracy drop on darker skin?
- What do apps miss in skin of color?
- How are labels and datasets changing?
- How can you compare an acne app?
Why does accuracy drop on darker skin?
The two largest public acne sets, ACNE04 and AcneSCU, contain only Han Chinese subjects. That finding comes from a 2025 systematic review in the Journal of Personalized Medicine. Models trained on narrow sets risk bias on other skin of color. Stanford researchers built 656 confirmed Diverse Dermatology Images across skin tones.
Leading models lost 29-40% ROC-AUC, a score for diagnostic accuracy, versus original results, per Science Advances Stanford DDI study. Losses were worst on dark skin and uncommon diseases. A 2025 evaluation cited in a 2026 UCLA review found ChatGPT dermatology accuracy fell about 20% for people with skin of color. The authors linked the drop to training sets skewed toward lighter skin.
What do apps miss in skin of color?
Acne in skin of color more often leaves dark marks after healing. Doctors call this post-inflammatory hyperpigmentation, or PIH. It appeared in 80% of dark-phototype patients in one 339-patient study, per Contemporary Pediatrics.
Keloid risk is 5-16 times higher than in White skin, in the same report. Thick scars and lasting spots change treatment choices. An app trained on light skin may grade redness only and miss brown or purple marks.
How are labels and datasets changing?
The Fitzpatrick scale measures sun photosensitivity, not skin tone, per npj Digital Medicine. The 10-shade Monk scale and Pantone guide better capture tone diversity for labeling AI data.
Google and Stanford released the open-access SCIN dataset to broaden everyday-condition coverage. It holds 5,000+ volunteer contributions and 10,000+ images with self-reported and dermatologist-estimated tones. Tones include both Fitzpatrick and Monk labels, per Google and Hugging Face.
How can you compare an acne app?
Ask for proof of testing in your skin-tone range, not broad claims. Look for validation that names Fitzpatrick I-VI or Monk 1-10 results.
FDA now expects Diversity Action Plans for pivotal trials, per Medscape's summary of June 2024 draft guidance Medscape summary of FDA diversity guidance. If a vendor cannot show tone-specific results, treat its score as untested for darker skin. Save the app's tone-range statement and show it at your visit.
- validation included Fitzpatrick I-VI or Monk 1-10 results
- independent dermatologist labels checked the images
- FDA marketing status is listed clearly
- pivotal-trial plan addresses diversity enrollment
You Might Also Like
- Skin-Tone Equity Gaps in 2026 Acne AI Research: How People With Skin of Color Can Avoid AI Hype
- Skin-Tone Equity Gaps in 2026 Acne AI Research: Why Clinician Review Still Matters for People With Skin of Color
- Recovery and Pigment Safety in the 2026 Split-Face Acne Scar Trial: How People With Boxcar Acne Scars Can Avoid Overreading the Headline