Particle.news

Download on the App Store

AI Analysis Spots Subtle Facial Cues Tied to Mild Depression

The findings suggest a non-invasive screening route, with validation needed beyond a small, culturally specific sample.

Overview

  • The peer-reviewed study from Waseda University was published on August 21, 2025 in Scientific Reports by Eriko Sugimori and Mayu Yamaguchi.
  • Researchers analyzed short self-introduction videos from 64 Japanese undergraduates and had 63 peers rate expressivity and likability while OpenFace 2.0 tracked facial action units.
  • Students reporting subthreshold depressive symptoms were rated as less friendly, expressive, and likable yet were not judged more stiff, fake, or nervous.
  • AI identified more frequent inner brow raiser, upper lid raiser, lip stretcher, and mouth-opening actions in these students, with micro-movements linked to depression scores and largely invisible to untrained observers.
  • The authors highlight potential use for early, non-invasive screening in schools, universities, workplaces, and digital platforms while cautioning that cultural factors and the small sample limit generalizability.