New Insights into Early Eye-Tracking Biomarkers for Autism Spectrum Disorder Risk Assessment

A recent study shows that brief, dynamic social-attention paradigms and novel gaze-retention (GRI) metrics distinguish toddlers with higher versus lower ADOS-indicated risk—offering earlier, objective markers that could change referral timing and accelerate diagnostic pathways.

The observational cohort enrolled 74 toddlers aged 12–30 months from elevated-likelihood, near-clinical screening and referral settings and used ADOS scores as the benchmark. Investigators presented dynamic social-attention stimuli (preferential-looking and ostensive joint-attention scenes) and recorded gaze with a custom, calibration-free device, extracting AOI ratios, fixation measures, and gaze-retention variables. Pairing age-appropriate, attention-eliciting stimuli with a validated clinical standard in a near-clinical sample supports evaluation of early markers.

GRI metrics performed strongly alongside AOI ratio measures while short-duration fixation metrics showed more variable separation. Key outcomes included gaze-retention interval (GRI) density and duration, total gaze time, GRI-in-AOI ratios, AOI social/non-social gaze-time ratios, and conventional fixation measures. In the preferential paradigm the social AOI ratio correlated negatively with ADOS (r = –0.63) while the non-social ratio correlated positively (r = 0.59); reported Hedge’s g effect sizes for high- versus low-ADOS groups ranged approximately 0.5–1.19. Preferential-looking dynamic social scenes produced the strongest discrimination between higher and lower ADOS-indicated risk.

Evidence shows the gaze-retention interval approach is more robust with noisy toddler data because GRIs integrate gaze across broader spatial and temporal windows, tolerate brief tracking gaps, and thereby reduce data loss and subject exclusion. In this cohort the median missing gaze rate was 9.5% and only 7.5% of enrolled toddlers were excluded overall, while GRI-based variables produced large effect sizes (Hedge’s g up to ~1.19) and correlations with ADOS in the hrADOS subgroup that reached r ≈ 0.77 for some measures. GRI-based metrics therefore lower analytic exclusion, preserve statistical power, and show larger group separation than many fixation-only metrics—providing a practical advantage for clinical-age eye-tracking.

Clinically, objective eye-tracking thresholds could prompt expedited comprehensive diagnostic evaluation or referral to early-intervention services when a toddler’s GRI and AOI pattern deviates beyond empirically derived local norms; such use should follow replication and local validation before routine implementation. These objective markers are adjuncts that complement behavioral screening and parent history rather than replacements. Implementing short, standardized eye-tracking triage sessions could help prompt earlier diagnostic scheduling within local pathways.