Researchers at Shenzhen University have developed an AI system that detects subtle visual cortex errors linked to future depression in teenagers, offering potential for early intervention.
A deep-learning system trained on routine MRI scans may be able to identify teenagers at risk of depression years before symptoms emerge, according to researchers at Shenzhen University in a study published in Science Advances on 16 August. The work points to a more specific brain mechanism than earlier imaging efforts: not a broad structural change, but a subtle error in how the visual cortex processes angry facial expressions.
The researchers examined data from the long-running IMAGEN study, which followed adolescents in Europe with scans, blood tests and questionnaires at ages 16, 19 and 23. They found that young people who later developed depression showed weaker neural responses to angry faces. The model they built suggests this happens when top-down emotional expectations become too rigid, blurring rather than sharpening the brain’s reading of anger. In effect, the visual cortex appears to underrepresent a signal that should be especially salient.
The team then tested the marker prospectively in participants who had no clinically relevant emotional symptoms at age 19 and found it could predict depression by age 23. They also reported validation in a separate European clinical cohort, suggesting the signal was not confined to one dataset. According to the study, the brain signature was linked not only to symptom patterns but also to known genetic risk for depression, strengthening the case that it reflects underlying biology rather than a momentary psychological state.
That still does not make it a ready-made screening test. The cohorts were European adolescents, so the findings may not travel cleanly to other populations. The effect size was also small, which is common in large population studies but limits near-term clinical use. Even so, the result stands out in a field where many AI-based depression models have struggled to hold up outside the original sample.
Its wider significance may lie in treatment rather than prediction. The researchers argue that the visual cortex could be a practical intervention target, especially since earlier work has suggested repetitive transcranial magnetic stimulation in that region may ease depressive symptoms. For now, the study offers something psychiatry has long lacked: a mechanistic brain signal that appears before illness begins and can be measured, tested and potentially changed.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
