AI-enabled eye-movement and emerging multimodal frameworks for precision dyslexia screening and reading pattern analysis

Dyslexia is a neurodevelopmental disorder that impairs reading accuracy and fluency, and early identification is important for timely intervention initiation. Traditional assessment methods are time- and resource-intensive. The research analyzed 23 studies from 2015 to 2026 focused on using eye-movement tracking and machine learning for dyslexia screening. Dyslexic readers consistently exhibited longer fixation durations, increased regressive behavior, and reduced saccadic efficiency. Machine-learning algorithms achieved classification accuracy ranging from approximately 80 to 95 percent, with some studies reporting values approaching 99 percent under specific conditions. However, the research revealed significant heterogeneity in datasets and validation methods. Multimodal approaches combining gaze data with additional data sources were identified in only three of the 23 studies. Computational systems based on eye movements represent a promising and non-invasive method for scalable dyslexia screening.