The study, published in Nature Mental Health, analysed electronic health records from more than 140,000 children. Researchers trained a specialised AI model to review medical history from birth through early childhood. The model learned to recognise combinations of developmental, behavioural and clinical events that often appeared years before an ADHD diagnosis.
The tool was highly accurate at estimating future ADHD risk in children age 5 and older and performed consistently across sex, race, ethnicity and insurance status. The AI does not make a diagnosis; it flags children who may benefit from closer attention by a paediatric primary care provider or an earlier referral for ADHD assessment. The authors say earlier screening and support could improve outcomes, but they call for further studies before routine clinical use.
Difficult words
- analyse — examine information to find important factsanalysed
- electronic health record — digital file with a person's medical informationelectronic health records
- specialised — made for a particular job or task
- recognise — see patterns or important information in data
- developmental — connected to how children grow and learn
- behavioural — relating to a person's actions or habits
- diagnosis — a doctor's decision about a health problem
Tip: hover, focus or tap highlighted words in the article to see quick definitions while you read or listen.
Discussion questions
- Do you think earlier screening for ADHD could help children? Why?
- Would you want a paediatrician to use an AI tool that flags children for assessment? Why or why not?
Related articles
AI and Wearable Devices for Type 2 Diabetes
A meta-review from the University at Buffalo examines AI-enhanced wearable devices for Type 2 diabetes and prediabetes. The study finds predictive benefits and important limits, and calls for larger, more transparent studies before routine clinical use.
Exercise helps reduce 'chemo brain' during chemotherapy
Researchers tested a home exercise program during chemotherapy and found patients who followed it kept daily activity and reported clearer thinking. The trial compared a six-week exercise plan with standard care during treatment.
Older adults more exposed to low-credibility online health information
Researchers at the University of Utah tracked web activity of more than 1,000 US adults for four weeks. They found low-credibility health sites were rare overall but visits were concentrated among older adults, especially those who lean right politically.