Artificial intelligence models using eye-tracking technology can identify children with autism spectrum disorder (ASD) from typically developing peers with 85% accuracy, according to a 2026 meta-analysis.
Symptoms of ASD often appear before age three, with diagnoses possible by 18 months. Current clinical guidelines recommend screenings at 18 and 24 months.
Eye-tracking has emerged as a promising tool for objective autism detection. Machine learning algorithms analyze gaze patterns to differentiate behaviors linked with ASD.
Findings from over 2,300 participants across 25 studies showed an 85% pooled accuracy, with 86% sensitivity and specificity. Performance varied based on age, stimuli, and the algorithm used.
Despite strong results, researchers caution about limitations. Small sample sizes and inconsistent methodologies hinder broader application.
Experts call for standardized protocols and large-scale trials to validate these tools before integrating them into routine pediatric assessments.