UPSC Psychology & Ethics - упрощенные уроки

Can AI predict depression years before symptoms appear? A 2026 study published in Science Advances combined brain imaging + deep learning to identify adolescents who may be at greater risk of developing depression. Researchers analysed how adolescents' brains processed emotional facial expressions. They found that weaker neural representation of emotional signals—particularly a tendency towards negative perceptual bias—was associated with depressive symptoms. The interesting part: this neurocomputational signature was able to predict the onset of depressive symptoms up to four years later in a longitudinal adolescent sample. The research used: IMAGEN cohort: 1,332 adolescents Four-year follow-up prediction: 725 participants Independent clinical validation (STRATIFY): 411 participants It was also associated with genetic and polygenic risk for depression, suggesting a possible pathway: Genetic vulnerability → altered processing of emotional information → negative perceptual bias → greater depression risk This is a strong contemporary example of how the biological, cognitive and computational approaches can converge in explaining psychopathology. However, this should not be interpreted as AI being able to diagnose someone's future depression with certainty. It is a risk-prediction biomarker, and much more validation would be needed before such models could become routine clinical screening tools. A fascinating movement from simply treating psychological disorders after symptoms appear → towards identifying vulnerability before disorder develops.