Depression is one of the most common mental health conditions, yet its earliest signs are often easy to miss. A new study suggests artificial intelligence (AI) may help identify those signs by analyzing tiny, nearly invisible facial movements.
What the Study Found
Researchers at Waseda University in Japan asked 64 college students to record short self-introduction videos. Another group of students rated how expressive, friendly, and natural the speakers appeared. At the same time, an AI tool called OpenFace 2.0 tracked micro-movements in the students’ eyes and mouths.
The results revealed a clear pattern:
Students with subthreshold depression—a state of mild symptoms that doesn’t meet the criteria for clinical depression but can increase future risk—were seen as less friendly, expressive, and likable by their peers.
They did not, however, come across as stiff, fake, or nervous.
AI analysis showed specific movements, such as subtle eyebrow raises, eyelid shifts, and mouth muscle changes, were linked to higher depression scores.
Why It Matters
Subthreshold depression often flies under the radar because it doesn’t look like full clinical depression. Yet it can still affect daily life and increase the risk of developing more severe symptoms later on.
The researchers say that by picking up on faint cues invisible to the human eye, AI could help provide non-invasive, accessible mental health screening in schools, workplaces, and digital health platforms.
A Cultural Factor
The study focused on Japanese students, and the authors note that cultural norms shape how people express emotions. Future research will be needed to see whether the findings hold true in other countries and settings.
Looking Ahead
“Subtle non-verbal cues, such as facial expressions, shape social impressions and reflect mental health,” said study co-author Eriko Sugimori, PhD, associate professor at Waseda University. “Our approach provides a novel, accessible, and non-invasive tool for early detection of depression before clinical symptoms appear.”
The study was published in Scientific Reports on August 21, 2025.
Key Takeaways
Peer perception: Students with mild depression appeared less expressive and likable.
AI insights: Micro-movements in the eyes and mouth strongly correlated with depression scores.
Screening potential: The method could be applied in schools, workplaces, and digital platforms.
YOU MAY ALSO LIKE
