In an exciting advancement, researchers have successfully utilized deep learning AI to diagnose autism in children with 100% accuracy by screening retinal images. This innovative approach offers a non-invasive and objective method for early diagnosis, particularly valuable in areas with limited access to specialist child psychiatrists.
The study involved 958 participants, half with autism spectrum disorder (ASD) and half as age- and sex-matched controls. The AI algorithm, trained using retinal images and symptom severity test scores, demonstrated exceptional performance in differentiating between children with ASD and typical development, as well as assessing ASD symptom severity.
The researchers believe their AI-based model could serve as an objective screening tool for ASD from as young as four years old, addressing the urgent need for accessible and accurate assessments, especially in areas with restricted resources.
HEALTH
WORLD AFFAIRS
TECHNOLOGY
CREATIVE
