Summary: New research shows that artificial intelligence can accurately estimate a child's risk of developing ADHD years before a clinical diagnosis is made. By digging for “hidden patterns” in Electronic Health Records (EHR) from birth to childhood, AI identifies a combination of developmental and behavioral symptoms that human doctors might overlook during a short visit.
The tool is designed to act as a “clinical safety net,” ensuring that at-risk children receive early screening and support during critical developmental windows.
Important Facts
- Data Set: The researchers analyzed the medical history of more than 140,000 childrencreating a greater comparison base for those with and without ADHD.
- Early Detection: The AI model analyzes data from birth and becomes more accurate in estimating future risk with age 5before the estimated time of diagnosis.
- Equivalent Performance: An important finding was the consistent accuracy of the model across all demographics, incl gender, race, nationality, and insurance statussuggesting that it may help reduce disparities in ADHD care.
- Support, Not Diagnosis: The tool is clearly not an “AI doctor.” Its purpose is to flag children who should be prioritized for evaluation by primary care providers or specialists.
- Best results: Early detection is directly linked to improved health, social, and long-term outcomes, as it allows for evidence-based interventions before a child is born.
Source: Duke University
Attention-deficit/hyperactivity disorder (ADHD) affects millions of children, yet many go years without a diagnosis, missing the opportunity to receive early support that can change long-term outcomes even when early symptoms are present.
In a new study, Duke Health researchers found that artificial intelligence tools can analyze electronic health records to accurately estimate a child's risk of developing ADHD years before a typical diagnosis. By reviewing patterns in daily medical data, the method can help flag children who may benefit from early screening and follow-up.
The study, published in Natural Mental Health on April 27, highlights how powerful insights can emerge from information already collected during routine health care visits to aid early decision support by primary care providers.
“We have this incredibly rich source of information sitting in electronic health records,” said Elliot Hill, lead study author and data scientist in the Department of Biostatistics & Bioinformatics at Duke University School of Medicine.
“The idea was to see if the patterns hidden in that data could help us predict which children might later be diagnosed with ADHD, before that diagnosis usually happens.”
To reach these findings, researchers analyzed electronic health records from more than 140,000 children, with and without ADHD. They trained a special AI model to look at medical history from birth to childhood. The model learned to recognize a combination of developmental, behavioral, and clinical events that often occurred years before ADHD was diagnosed.
The model was highly accurate in estimating future ADHD risk in children age 5 and older, with consistent performance across patient characteristics such as gender, race, ethnicity, and insurance status.
Importantly, the tool does not make a diagnosis. It identifies children who may benefit from closer attention by their primary pediatric care provider or early referral for ADHD screening by a specialist.
“This is not an AI doctor,” said Matthew Engelhard, MD, Ph.D., in Duke's Department of Biostatistics & Bioinformatics, and senior author of the study. “It's a tool to help doctors focus their time and resources, so that children who need help don't fall through the cracks or wait years for answers.”
The researchers note that earlier screening tests can lead to earlier diagnosis and therefore earlier support, which has been linked to better academic, social, and health outcomes for children with ADHD. They also emphasize the need for more studies before such tools can be used in clinical settings.
“Children with ADHD can really struggle when their needs are not understood and adequate support is not available,” said study author Naomi Davis, Ph.D., associate professor in the Department of Psychology and Behavioral Sciences. “Connecting families with timely, evidence-based interventions is critical to helping them reach their goals and lay the foundation for future success.”
Hill and Engelhard also researched the use of AI models in predicting the potential risks and causes of mental illness in youth.
In addition to Hill Engelhard, and Davis, the authors of this study include De Rong Loh, Benjamin A. Goldstein, and Geraldine Dawson.
Sponsorship: The research was supported by grants from the National Institute of Mental Health (K01-MH127309, UL1 TR002553) and the National Center for Advancing Translational Sciences.
Important Questions Answered:
A: AI looks for time and combination of events, such as specific developmental delays, sleep disturbances, or frequent visits to behavioral concerns, which may seem insignificant on their own but together form the “risk signature” of ADHD.
A: No. Senior author Dr. Matthew Engelhard emphasizes that this is a resource management tool. It helps pediatricians know which children need more attention so they don't “fall through the cracks” while waiting years for a routine check-up.
A: Primary care providers often have limited time with patients. AI can scan thousands of pages of a child's medical history in seconds, highlighting relevant clinical trends that may have occurred years ago or with a different doctor.
Editor's Notes:
- This article was edited by a Neuroscience News editor.
- The journal paper is fully revised.
- More content has been added by our staff.
About this AI and ADHD research news
Author: Stephanie Lopez
Source: Duke University
Contact person: Stephanie Lopez – Duke University
Image: Image posted in Neuroscience News
Actual research: Closed access.
“Fetal and post-natal metal metabolism-related changes in brain function are associated with behavioral deficits in children” by Elliot D. Hill, De Rong Loh, Naomi O. Davis, Benjamin A. Goldstein, Geraldine Dawson and Matthew Engelhard. Natural Mental Health
DOI:10.1038/s44220-026-00628-2
Abstract
Attention Deficit Hyperactivity Disorder (ADHD) is a common neurodevelopmental condition that can have a negative impact on long-term outcomes for individuals. Early diagnosis is important, but demographic and treatment differences can delay diagnosis.
Using electronic health records (EHRs) from a cohort of over 720,000 patients, we pre-trained a baseline EHR model. We then fine-tuned it to predict the probability of ADHD diagnosis and time from birth to age 9 in a pediatric cohort of over 140,000 patients.
At 5 years of age, the model achieved a time-dependent area under the receiver operating characteristic curve of 0.92 over a 4-year horizon. Overall, the model retained its validity across patients with different demographics, including gender, race, ethnicity and insurance status.
Our factor significance analysis found that ADHD was significantly associated with developmental, behavioral and psychological conditions. Our results suggest that EHR-based predictive models can help providers reliably identify children with ADHD in a timely manner.
