Feed

Adolescent depression and anxiety are common, heterogeneous, and often recognized late. Young people may present with sleep disturbance, irritability, somatic complaints, school refusal, social withdrawal, attentional difficulties, or functional decline before they meet clear diagnostic thresholds. Artificial intelligence has been proposed as a way to improve early identification, monitoring, triage, and treatment planning. However, predictive accuracy alone does not establish clinical value. This Mini Review examines what artificial intelligence can realistically add to the care of adolescents with depression and anxiety, with emphasis on decision support rather than automated diagnosis. Current evidence suggests potential value in prospective risk prediction, digital phenotyping, measurement-based monitoring, conversational interfaces, and early detection of poor treatment response. These tools may help clinicians identify young people who need closer follow-up, structured assessment, safety planning, or treatment adjustment. However, the evidence remains limited by retrospective designs, selected samples, inconsistent outcome definitions, weak external validation, limited calibration, uncertain incremental value over standard clinical assessment, and insufficient testing in real workflows. Digital and conversational tools also raise concerns about privacy, consent, safeguarding, equity, and over-monitoring. Clinically useful artificial intelligence should therefore be evaluated by whether it improves concrete decisions: who is assessed earlier, monitored more closely, escalated for safety concerns, or offered a different treatment pathway. Future research should prioritize measurement precision, cultural transportability, realistic comparators, and evidence that model-guided care improves outcomes for young people.