Researchers at Duke Health have developed an artificial intelligence model that may detect attention-deficit/hyperactivity disorder (ADHD) in children years before they are formally diagnosed, according to a study published in the journal Nature Mental Health on April 27, 2026.
The AI tool analyzed functional MRI (fMRI) brain scans from 1,200 children aged 9 to 10, identifying patterns of brain connectivity associated with ADHD. The model predicted which children would receive an ADHD diagnosis up to two years later, with an accuracy of about 80%.
Lead author Dr. John Smith, a psychiatrist at Duke University Medical Center, said: 'Early identification could allow for interventions before symptoms become severe, potentially improving long-term outcomes.' The study was funded by the National Institute of Mental Health.
