An Automatic AI-Based Algorithm That Grades the Scalp Surface Exfoliating Process From Video Imaging: Application to Dandruff Severity and Its Validation on Subjects of Different Ages and Ethnicities

The study presents an AI-based algorithm using the ScalpConsult Pro device to assess dandruff severity through high-resolution scalp imaging. The algorithm was trained on 3,600 images and validated on 460 images from 234 subjects aged 18-60 across diverse ethnicities in France, China, and Mauritius. It showed a high correlation (r² = 0.952, p < 0.001) with expert evaluations, with a Mean Absolute Error mostly below 0.3 grading units. The study highlights the device's accuracy and efficiency over traditional methods, though it notes higher errors in the Mauritius cohort. Future research aims to expand the dataset to over 20,000 subjects to improve generalization across different ethnicities and phototypes. The device also shows potential for diagnosing other scalp and skin conditions.