Duke study: AI model detects ADHD in kids as young as 5
Add Axios as your preferred source to
see more of our stories on Google.

Illustration: Maura Losch/Axios
Duke University researchers created an artificial intelligence model they say can accurately detect ADHD in 5-year-olds' health data.
Why it matters: That's years earlier than most kids are diagnosed, an example of how AI tools that analyze large datasets could reshape health care in the coming decades.
Driving the news: A paper published Monday by Nature Mental Health showed the new AI model for detecting attention-deficit/hyperactivity disorder works by age 5.
- That is "well before that diagnosis usually happens," lead author Elliot Hill, a data scientist in the Duke University School of Medicine, said in a news release.
How it works: Researchers fed the data from 140,000 children (birth to age 9) to an AI model already trained on the electronic health records of more than 720,000 people.
- The AI model picked up on hidden patterns in the data and could accurately predict a diagnosis across demographics like insurance status, sex and race, the paper says.
What they're saying: "This is not an AI doctor," study co-author Matthew Engelhard, a fellow Duke data scientist, said in the release. "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."
- Naomi Davis, a co-author from the field of psychiatry and behavioral sciences, added that "children with ADHD can really struggle when their needs aren't understood and adequate supports are not in place."
By the numbers: Around 11% of American children are diagnosed with ADHD, according to 2022 data analyzed by the Centers for Disease Control and Prevention.
- That's 15% of boys and 8% of girls.
The intrigue: Just over half (54%) of U.S. children diagnosed with ADHD were taking medication, per CDC data.
- Meanwhile, demand for ADHD medications has surged in recent years among adults, causing drug shortages at times.
What's next: The researchers say the model, while promising, should undergo further study before it can be used in the real world.
