Let's talk about the elephant in the room: most skin analysis AI is biased against darker skin tones. And we're here to fix that.
The Problem: Training Data Bias
Most dermatology AI models are trained on datasets where 80%+ of images feature Fitzpatrick types I-III (lighter skin). This means the AI literally doesn't know what "healthy" looks like on darker skin.
Common failures include:
- Inflated redness scores — The AI interprets natural melanin warmth as "redness" or "irritation"
- Penalized glow scores — Darker skin gets lower scores simply because the AI associates brightness with health
- Missed conditions — Conditions like rosacea, eczema, and skin cancer present differently on melanated skin and are frequently missed
- Ignored PIH — Post-inflammatory hyperpigmentation, the #1 skin concern for MST 6-10, is often not even tracked
Our Solution: Melanin Intelligence™
GlowLog uses the Monk Skin Tone (MST) scale — a 10-point scale developed by Dr. Ellis Monk at Harvard specifically to represent the full range of human skin tones.
When you enable Melanin Mode:
- The AI calibrates ALL metrics for your specific MST level
- Redness detection shifts from color-based to texture-based analysis for MST 6+
- PIH tracking activates — monitoring dark spots over weeks and months
- The Ashy Index measures visible dryness on darker skin tones
- Glow Score is never penalized for natural melanin darkness
- Product recommendations exclude ingredients known to cause PIH in melanated skin
The Black Skin Directory
We've also partnered our recommendations with the Black Skin Directory, making it easy to find dermatologists who specialize in melanated skin care.