Incorporating eye-tracking biomarkers into pediatric autism assessments may make identifying the condition easier, according to new findings published in JAMA Network Open.
Researchers created an artificial intelligence–based tool to help primary care clinicians and pediatricians spot potential cases of the neurological condition, according to Brandon Keehn, PhD, associate professor in the Department of Speech, Language, and Hearing Sciences at Purdue University in West Lafayette, Indiana, and an author of the study.
Most primary care clinicians do not receive specialized training in identifying autism, and around a third diagnose the condition with uncertainty, according to Dr. Keehn. The tool helps clinicians by incorporating their diagnosis and self-reported level of certainty with eye-tracking biomarkers. A clinical psychologist also assessed children, either verifying or confuting the earlier results.
The tool produced the same diagnosis as that from a psychologist in 90% of cases. When children were assessed using eye biomarkers alone, the diagnosis was aligned with that of a psychologist 77% of the time.
“This is the first step in demonstrating both that eye-tracking biomarkers are sensitive to autism and whether or not these biomarkers provide extra clinical information for primary care physicians to more accurately diagnose autism,” Dr. Keehn told this news organization.
The study took place between 2019 and 2022 and included 146 children between 14 and 48 months old who were treated at seven primary care practices in Indiana. Dr. Keehn and colleagues asked primary care clinicians to rate their level of certainty in their diagnosis.
During the biomarker test, toddlers watched cartoons while researchers tracked their eye movements. Six biomarkers included in the test were based on previous research linking eye movements to autism, according to Dr. Keehn.
These included whether toddlers looked more at images of people or geometric patterns and the speed and size of pupil dilation when exposed to bright light.
Most toddlers produced a positive result for autism in only one biomarker test. Dr. Keehn said this confirms that children should be tested for a variety of biomarkers because each patient’s condition manifests differently.
Dr. Keehn said his team is still a few steps away from determining how the model would work in a real clinical setting and that they are planning more research with a larger study population.
Alice Kuo, MD, a pediatrician specializing in autism at the University of California, Los Angeles (UCLA), said primary care clinicians should feel comfortable making an autism diagnosis.
“Any tool that helps them to do that can be useful, since wait times for a specialist can take years,” Dr. Kuo, also the director of the Autism Intervention Research Network on Physical Health at UCLA, said.
However, Dr. Kuo said she is concerned about the cases that were falsely identified as positive or negative.
“To be told your kid is autistic when he’s not, or to be told your kid is not when he clinically is, has huge ramifications,” she said.
The study was funded by the National Institute of Mental Health, the Riley Children’s Foundation, and the Indiana Clinical and Translational Sciences Institute. Dr. Keehn reported payments for workshops on the use of the Autism Diagnostic Observation Schedule.
A version of this article appeared on Medscape.com .