Researchers detected ADHD in children with 96.9% accuracy by analyzing fundus images using artificial intelligence.
South Korean scientists have developed an artificial intelligence system that can identify ADHD in children using retinal images. The noninvasive model achieved 96.9% accuracy and could be incorporated into healthcare centers.
An image of the eye could become a key tool for detecting ADHD.
A Korean study found that artificial intelligence can detect ADHD with nearly 97% accuracy using fundus images in children.
Childhood ADHD, ocular diagnosis, artificial intelligence in health, fundus, neurodevelopment, retina, AI model, South Korea
A research team in South Korea has developed an innovative method for diagnosing attention-deficit/hyperactivity disorder (ADHD) in children by analyzing images of the ocular fundus. The artificial intelligence-based system achieved 96.9% accuracy using only retinal data.
The study, published in the scientific journal Nature , included 646 children, half with a confirmed diagnosis of ADHD and the other half with no history, selected by similar age and sex. By analyzing the shape, density, and thickness of retinal blood vessels, as well as optic disc characteristics, the algorithm was trained to identify patterns associated with the disorder.
The specialists explained that these variables not only made it possible to distinguish between children with and without ADHD, but also to detect alterations linked to selective visual attention, a function frequently affected by this type of disorder.
This development is part of a growing trend in medicine seeking new visual and noninvasive biomarkers for diagnosing neurological diseases, with the goal of reducing dependence on more complex methods such as extensive clinical interviews, cognitive tests, or lengthy evaluations by specialists.
The researchers emphasized that this procedure is quick, requires no sedation, and poses no risks to the patient. It could be easily integrated into primary care, child and adolescent mental health services, or pediatric consultations.
In countries like Spain, where the Spanish Association of Pediatrics estimates that nearly half a million children have ADHD, the availability of a rapid and objective screening tool could make a significant difference, especially in areas with high demand and limited resources.
However, the authors also highlight certain limitations of the study. For example, children diagnosed with autism were not included, which limits the system's applicability in cases with comorbidity. Furthermore, the average age of the sample was 9.5 years, so its effectiveness in older adolescents or adults has not yet been tested.
The team is already working to expand the research to other age groups and explore whether this technology could also be used to detect other neurodevelopmental disorders or even psychiatric illnesses.
Finally, they warn that while the method has great potential, its implementation will require an appropriate regulatory framework that guarantees patient privacy, clinical validation of the system, and constant medical supervision.
For now, the tool does not replace traditional diagnosis, but it is positioned as a valuable complement that could improve the speed and accuracy of identifying ADHD in children.
