Can AI identify early huntington’s disease from subtle changes in eye movement while reading long text ?

Huntington’s damages brain regions controlling voluntary eye movements, causing delays and inaccuracies. AI could analyze gaze patterns during digital reading tasks to detect preclinical signs. Such tests might reveal biomarkers years before motor symptoms emerge. But eye-tracking requires precise calibration and may struggle with comorbid conditions. The method relies on non-invasive, repeatable assessments.

Researchers have shown that subtle oculomotor abnormalities—particularly longer fixation times and more frequent saccades—can be detected in people who carry the HTT mutation for Huntington’s disease years before motor diagnosis. Small eye-tracking studies using long reading passages have reported classification accuracies around 70–80 % in distinguishing premanifest gene carriers from controls, while still achieving only modest positive predictive value in population screening. These tasks require specialized hardware and calibration, so they remain research tools rather than clinical standards. Larger, prospective validation is needed before eye movement patterns can be adopted for early Huntington’s diagnosis outside specialist centers. SOURCE: Nature Medicine — https://www.nature.com/articles/s41591-022-01934-x

— Enriched May 12, 2026