Identifying reading disabilities through eye movements: a validation study using Lexplore and AI-driven technology

This study explores the potential of eye-tracking technology combined with artificial intelligence (AI) to identify early reading disabilities (RD). Findings indicate significant differences in fixation patterns between elementary students with and without RD, suggesting eye-tracking as a promising screening tool for classroom use. Students’ reading failure is a systemic problem across the United States. Early reading interventions show promise in addressing reading disabilities (RD); however, traditional measures are costly, time-consuming, and unreliable. Recent researchers conclude that AI tools, such as eye-tracking, could detect RD for earlier intervention. This study investigated the potential impact of eye-tracking data during a Lexplore reading screening to determine if differences existed between the (a) average fixation time and (b) proportions of fixations to total stimuli duration while reading with 12 elementary students with RD and 17 students without RD. Although Lexplore incorporates AI and previously trained machine learning algorithms, the present study's goals were not to train, modify, or validate AI models. Instead, researchers analyzed pre-generated eye tracking output using traditional statistical methods to examine group differences regarding reading difficulties. This study showed statistically significant differences between groups in both average and proportions of eye fixations. The findings from this exploratory study indicate a need for further investigations regarding eye-tracking devices to screen, identify, and monitor elementary students potentially at risk for RD.