Autism in Young Children: Eye-Tracking Technology May Improve Diagnosis

A composite eye-tracking biomarker may be useful in the clinical diagnosis of autism in young children in primary care settings, according to study results published in JAMA Network Open.

Previous research has shown the benefits of eye tracking in identifying early autism diagnostic biomarkers.

Researchers of a prospective study aimed to determine the effect of eye-tracking biomarkers on distinguishing young children with and without autism, as well as whether a combination of methods — eye tracking and primary care practitioner (PCP) diagnosis — can improve diagnostic outcomes.

Eligible participants, aged between 14 and 48 months, were enrolled in the Early Autism Evaluation (EAE) Hub system that was set up at primary care practices.

The researchers collected participant data from caregivers through an electronic survey, following which an autism diagnostic assessment by PCPs and eye-tracking biomarker battery (index test) were conducted. The PCP diagnoses for autism and diagnostic certainty were rated on a 5-point Likert scale, ranging from completely certain to not at all certain. For the eye-tracking biomarker battery, drift check and correction were completed based on a series of tests.  

Primary study outcomes were sensitivity and specificity of the eye-tracking test, which was based on certain quality indices; secondary outcomes were sensitivity and specificity of an integrated approach — the index test and PCP diagnosis.

Between 2019 and 2022, 146 children were eligible in the analysis. Of these, 102 (boys, 76%; mean age, 2.6 years) were diagnosed with autism and 44 (boys, 59%; mean age, 2.4 years) were not. The majority of the study population was of non-Hispanic White race and ethnicity.

Six eye-tracking measures, including nonsocial preference, no-shift gap effect, pupillary light reflex (PLR) latency and amplitude, and resting and exploratory fixation duration were associated with autism diagnostic outcomes. Specifically, nonsocial preference was associated with decreased scores on the early learning and adaptive behaviors scale (r=-0.26 and r=-0.31) in the autism vs nonautism group.

Of note, patients with a reference diagnosis of autism were identified using a single, unique eye-tracking biomarker.

Diagnostic outcomes were found to be concordant with the reference diagnosis for 113 of the 146 (77%) participants. Sensitivity and specificity for autism and nonautism diagnoses were noted to be 77.5% and 77.3%, respectively.

The researchers observed that the strongest predictor of reference autism outcomes was the composite biomarker; 89% of those who had a biomarker-positive score had an autism outcome.

Consistent outcomes were seen in diagnoses by PCPs vs standard reference in 114 of 127 (90%) participants, with a significant diagnostic agreement (sensitivity 90.7% and specificity 86.7%).

Overall, the researchers concluded, “The findings of this diagnostic study suggest that multiple eye-tracking indices may be sensitive to autism, provide additional information beyond PCP diagnostic outcome and certainty, and, when integrated with these measures, may facilitate more accurate autism diagnosis.”