Lexplore — AI Eye-Tracking Screening for Dyslexia and Reading Difficulties

Lexplore was spun out of Sweden's Karolinska Institutet in 2016 to commercialise eye-tracking research into an AI-scored reading assessment. A child reads a short passage while a camera tracks eye fixations and regressions; a machine-learning model compares the pattern against a normed reference set to flag dyslexia and reading-difficulty risk in about five minutes, without requiring a specialist to administer it.

The company reports having screened roughly 8,000 children across Sweden and the US and describes its own accuracy as 'an impressive 95%' — a company claim, not independently audited. In May 2026, researchers from the University of Central Florida and the University of Kansas published the first independent peer-reviewed validation in Frontiers in Education: studying 29 elementary students (12 with a reading disability, 17 without) at a single charter school, they found a statistically significant, very large difference in average fixation time between the two groups (t(27)=6.696, p<.001, Cohen's d≈2.64). The same study noted the sample was small and single-school, that 16 additional 'emerging readers' produced no usable eye-tracking data, and criticised the proprietary algorithm for lacking transparency about how it converts eye movements into a risk score.