Stanford-led study finds three brain patterns in children with ADHD, anxiety or both, and they show up before diagnosis
AI sorted thousands of brain scans into three types that don’t line up neatly with today’s diagnoses. Researchers hope they someday help match kids to treatment faster.
Children with attention-deficit/hyperactivity disorder (ADHD), anxiety or both fall into three distinct brain patterns, and those patterns don’t follow the diagnostic labels doctors use, according to a Stanford Medicine-led study announced Nov. 10.
ADHD and anxiety are two of the most common mental health conditions in childhood, and they often overlap. Doctors diagnose both with symptom checklists, not brain tests. That makes it hard to know why a child struggles, and finding a medicine that works can take weeks or months of trial and error, said co-senior author Yu Zhang, a Stanford assistant professor of psychiatry and behavioral sciences.
What the team did
The researchers used data from the Adolescent Brain Cognitive Development study, a long-term study of brain development that enrolled more than 11,000 U.S. children at age 9 or 10. They studied 3,508 children with a diagnosis, 693 who were diagnosed later and 1,947 who never were. All had fMRI scans, which track blood flow to show which brain areas work together. An AI model compared those patterns with the children’s behavior. The team checked its results on scans of 224 more children from a separate study.
The three types
- Externalizing-disinhibited: more defiance, impulsivity and rule-breaking, and more often an ADHD diagnosis.
- Internalizing-distressed: more guilt, tension and perfectionism, and more often an anxiety diagnosis.
- Mixed: more attention problems, irritability and self-injury, and more often ADHD.
Every type appeared in every diagnosis group. The patterns were also visible in early scans of children who were diagnosed only later. “We may be able to use the biotypes as an early indicator of these conditions,” Zhang said.
The biotypes are not a test parents can request. Zhang said they probably won’t replace symptom-based diagnosis but could help future clinical trials learn which treatments work for which pattern. The study was published Oct. 16 in Nature Mental Health and was funded mainly by the National Institutes of Health.