TRANSFORM-MH: Leveraging Longitudinal Data to Transform Early Intervention in Adolescent Mental Health

TRANSFORM-MH unites advanced longitudinal data science and experimental research to inform early-intervention in adolescent mental health. We focus on age 10–11 years, a key school transition point, to identify children at high-risk of depression and anxiety before problems escalate using rich longitudinal data from 16 datasets from Chile, Denmark, Ethiopia, France, India, Norway, Peru, UK, USA, Uruguay, and Vietnam. We will apply machine learning, causal inference, and Mendelian randomisation to identify early-life predictors and build robust predictive risk-scores for individual children. In parallel, we will co-design a practical, acceptable and affordable intervention with children and adolescents, families, schools and services to integrate these scores into school systems, aiming to support children most at risk. We will test this intervention in a cluster randomised trial across 50 Bradford schools, evaluating its impact on access to support, mental health outcomes and potential harms. We will use linked primary and secondary care, school, and local-authority data to evaluate longer-term outcomes to age 18 and beyond. We will harness established links to national policymakers to promote implementation. Overall, TRANSFORM-MH aims to develop innovative, interpretable prediction models, generate strong evidence for earlier, better-targeted mental health support for those children who need it most.