Eye Tracking Biomarkers May Improve Autism Diagnoses

A recent paper published in JAMA Network Open describes a research study of autism diagnosis using eye-tracking biomarkers at primary care clinics in Indiana. Researchers traveled to practices participating in the Indiana Early Autism Evaluation Hub system, conducting blind research-grade evaluations of 146 children aged 14-48 months.

Eye-tracking biomarkers that measure social and nonsocial attention and brain function differentiate young children diagnosed with autism from those with other neurodevelopmental disabilities. To date, there’s been a gap in translating the biomarkers so that they would be useful.

In this study, children sat in a highchair or caregiver’s lap and watched videos on a computer screen while their eye movements and pupil size were recorded. When clinician diagnosis and diagnostic certainty were combined with eye-tracking biomarker metrics, the model’s sensitivity was 91%, and specificity was 87%—the autism diagnosis was more accurate.

Next, the team plans to use artificial intelligence to conduct a large-scale replication and validation study of their diagnostic model. Then, they will conduct a clinical trial to study its effectiveness in real-time primary care evaluations.