WASHINGTON (dpa-AFX) - New research suggests that even mild depression symptoms below the threshold of a clinical diagnosis can subtly alter facial expressions in ways detectable by both people and artificial intelligence.
The study, published in Scientific Reports by researchers at Waseda University in Japan, highlights how AI-driven analysis of short videos could provide an early-warning tool for depression.
The team recruited 64 Japanese university students, dividing them into two groups: those with minimal symptoms and those with 'subthreshold depression,' or mild symptoms. Each student recorded a brief self-introduction video.
A separate group of 63 peers then rated the clips on traits such as friendliness, naturalness, and likeability. Students with mild depression symptoms consistently scored lower on all positive traits, though they were not perceived as more stiff, fake, or nervous than their healthier peers-suggesting muted positive expressivity rather than overt negativity.
To complement human perception, the researchers used OpenFace 2.0, an open-source AI tool, to track facial muscle movements. The analysis revealed distinct patterns in the subthreshold group, including heightened use of the inner brow raiser, lip stretcher, and jaw drop micro-expressions often linked to tension or discomfort.
These movements correlated with higher depression scores, indicating measurable, nonverbal cues of early psychological distress.
The findings are limited by the use of self-report questionnaires and a culturally specific sample, but they open new possibilities for early detection.
Lead author Eriko Sugimori said the approach could be adapted for schools, workplaces, or digital health platforms.
'Our study provides a non-invasive AI-based tool for early detection of depression, enabling timely intervention and care,' she noted.
Copyright(c) 2025 RTTNews.com. All Rights Reserved
Copyright RTT News/dpa-AFX
© 2025 AFX News
