Autism Diagnosis Enhanced by Eye Tracking

Researchers examined the accuracy of eye-tracking biological markers in differentiating between autistic and non-autistic children during clinical examinations in community-based settings for Autism diagnosis. The study, published in JAMA Network Open, also investigated whether better diagnostic outcomes might be achieved by combining biological markers with PCP diagnoses and diagnosis certainty.

Context

The gaps in autism diagnosis are worse among children from racial and ethnic minorities and impoverished neighborhoods. The high number of chilTo addresses the challenge. New community-based care delivery models are under development that combine clinical and biobehavioral techniques to boost early diagnostic accuracy and timeliness. Eye-tracking biological markers, which are non-invasive, low-cost, and feasible, offer promise for discovering early autism diagnostic biomarkers.

About the study

In the present prospective diagnostic investigation, researchers explored the reliability of eye-tracker biological markers employed in primary care clinical evaluations to detect autistic children in community settings. They determined if integrating these biomarkers with PCP diagnoses would increase diagnostic accuracy. Dren who need assessments—more than there are specialists—is the cause of the lengthy wait times for exams. Delays in diagnosis prevent early, evidence-based treatment, which reduces long-term care costs.

To address the challenge, new community-based care delivery models are under development that combine clinical and biobehavioral techniques to boost early diagnostic accuracy and timeliness. Eye-tracking biological markers, which are non-invasive, low-cost, and feasible, offer promise for discovering early autism diagnostic biomarkers.

The Early Autism Evaluation (EAE) PCPs proposed a sequential pediatric sample for a blinded ophthalmologic monitoring index assessment and expert evaluation during follow-up between June 7, 2019 and September 23, 2022. Seven EAE Hub centers referred 146 participants, ages 14 to 48 months, for the study. 146 of the 154 children who took part in the study generated sufficient data for at least one eye-tracking statistic.

Members of the study collected EAE Hub PCP data, and child caretakers filled out internet surveys. The team conducted an eye-tracking biomarker battery test and a follow-up criterion-standard autism diagnosis evaluation within 16 weeks following the EAE Hub examination.

The diagnosis was made by a licensed clinical psychologist using information from the Mullen Scales of Early Learning (MSEL), Vineland Adaptive Behavior Scale, Third Edition (VABS-3), Autism Diagnostic Observation Schedule, Second Edition (ADOS-2), and a caregiver interview. The PCPs at the EAE Hub provide diagnostic certainty and categorical diagnoses, with two categories of confidence: certain and uncertain (slightly, somewhat, or not at all certain).

The main study results were the eye-tracking index test’s specificity and sensitivity, as well as the composite measure’s substantial eye tracking indices when compared to clinical psychologists’ standard reference diagnosis of autism. The specificity and sensitivity of the combined approach, which comprised the index test, PCP diagnoses, and diagnostic certainty, were secondary outcomes.

The non-social preference, attentional disengagement, pupillary light reflex (PLR) latency and amplitude, tonic pupil size, oculomotor metrics, and passive visual exploration were all measured by the researchers using five eye-tracking biomarker batteries. The five paradigms in the battery were the following: GeoPreference, gap-overlap, PLR test, eye-tracking task while at rest, and passive visual exploration task. To find the best predictors for the reference standard autism diagnosis, the researchers employed binary logistic regressions, Pearson correlations, and a categorization and regression tree (CART) analysis.

Outcomes

The average age of research participants was 2.6 years; males made up 71% (n=104); Hispanics and Latinos made up 14% (n=21); and non-Latinos made up 66% (n=96). With 78% sensitivity and 77% specificity, 70% of the sample (n = 102) had a standard reference autism diagnosis, and 77% of the sample (n = 113) had autistic outcomes that were in line with the biomarker composite (index) and standard reference endpoints. Utilizing the composite biomarkers of the index test, diagnoses from primary care physicians, and certainty, 90% of the subjects (114 out of 127) had results that were in agreement with the reference, with 87% specificity and 91% sensitivity.

Reference standard autism outcomes, such as non-social preference, no-shift percentage, PLR latency and amplitude, and resting and exploratory fixation lengths, were found to be associated with six eye-tracking indices. A higher percentage of non-social preferences was associated with poorer MSEL and VABS-3 scores for the autism reference group, but not for the non-autism group, according to correlational analyses of relevant biomarkers and autism severity, developmental levels, and adaptive skills.

Every biomarker was unrelated, with the exception of the same concepts examined in two tasks (fixation time). Our model performed better than the other two models discovered, as evidenced by its mean cross-validation AUC of 0.90 and mean area under the receiver operating characteristic curve (AUC) of 0.93 for the three training runs that led to its selection. With a high mean cross-validated AUC of 0.90, the out-of-sample performance was good.

In summary

According to the study’s findings, in areas with a dearth of neurodevelopmental specialists, a multimethod approach to early autism diagnosis may improve access to trustworthy diagnoses. A number of eye-tracking indices may be sensitive to autism and offer details beyond the PCP diagnosis’s certainty and outcome.

The composite eye-tracking biomarker showed 87% specificity and 91% sensitivity when paired with the diagnosis and certainty of a primary care physician. It was linked to the best-estimate clinical diagnosis of autism.

The results suggest that offering primary care doctors a multimethod diagnostic approach could significantly improve the availability of prompt and precise autism diagnoses.

For more information: Eye-Tracking Biomarkers and Autism Diagnosis in Primary Care, Jama Network, doi:10.1001/jamanetworkopen.2024.11190