A new study from Duke Health suggests that artificial intelligence could help identify children at risk of attention-deficit/hyperactivity disorder (ADHD) years before they receive a formal diagnosis. The research, published in Nature Mental Health, demonstrates that AI tools can analyze routine electronic health records to estimate a child's likelihood of developing the condition, offering a pathway to earlier support.
ADHD affects an estimated 8 percent of children and teenagers globally, with symptoms including difficulty focusing, restlessness, and impulsivity. Many children go undiagnosed for years, missing opportunities for early intervention even when warning signs are present. The Duke team aimed to leverage the vast amount of data already collected in medical records to address this gap.
How the AI model works
The researchers analyzed health records from more than 140,000 children, both with and without ADHD, training an AI model to detect patterns from birth through early childhood. The system learned to recognize combinations of developmental, behavioral, and clinical events that often appeared years before an ADHD diagnosis. It proved highly accurate at estimating risk among children aged five and older, with consistent results across factors such as sex, race, ethnicity, and insurance status.
“We have this incredibly rich source of information sitting in electronic health records,” said Elliot Hill, lead author of the study and data scientist in the Department of Biostatistics & Bioinformatics at Duke University School of Medicine. “The idea was to see whether patterns hidden in that data could help us predict which children might later be diagnosed with ADHD, well before that diagnosis usually happens.”
Experts say earlier identification could lead to earlier diagnosis and support, which is linked to improved academic, social, and health outcomes for children with ADHD. “Children with ADHD can really struggle when their needs aren’t understood and adequate supports are not in place,” said Naomi Davis, an associate professor in the Department of Psychiatry and Behavioral Sciences and an author of the study. “Connecting families with timely, evidence-based interventions is essential for helping them achieve their goals and laying a foundation for future success.”
The tool is not designed to replace doctors or provide a complete diagnosis. “This is not an AI doctor,” said Matthew Engelhard from Duke’s Department of Biostatistics & Bioinformatics, and senior author of the study. “It’s a tool to help clinicians focus their time and resources, so kids who need help don’t fall through the cracks or wait years for answers.”
Similar AI approaches are also being explored to better understand risks and causes of mental illness in adolescents. In Europe, where healthcare systems vary widely from the UK’s National Health Service (NHS) to Germany’s statutory insurance model, such tools could help standardize early detection. For instance, the NHS notes that common ADHD symptoms in children include being easily distracted, struggling to listen, forgetting everyday tasks, and showing high levels of energy, such as fidgeting or tapping hands and feet.
While the study is based on US data, its implications are relevant for European health systems grappling with rising mental health demands. The ability to predict ADHD years in advance could reduce diagnostic delays, which are particularly acute in countries with limited child psychiatry resources, such as parts of Eastern Europe or rural areas in Spain and Italy. As synthetic drugs reshape global markets, straining Europe's health systems, innovative tools like this could help allocate scarce resources more effectively.
The research also highlights the potential of AI in pediatric mental health, an area where Europe has seen growing investment. For example, a recent study on urban vs rural upbringing shapes distinct mental health profiles in children underscores the need for tailored approaches. The Duke team’s work suggests that AI could be a key part of that toolkit, provided it is integrated carefully into clinical workflows.
As the technology matures, European regulators and health authorities will need to consider data privacy, algorithmic bias, and integration with existing systems. But for now, the study offers a promising glimpse into how routine data can be repurposed to improve children’s lives.
