Thompson Center experts test video screening tool for autism

COLUMBIA, MO (March 5, 2017) — When a family, doctor or teacher suspects a child has autism, they look to diagnostic experts to determine the diagnosis.

Waiting for a diagnostic appointment, though, can be agonizing – and long. Across the U.S., the average wait time for a diagnostic appointment for autism is 13 months, and locally, the University of Missouri’s Thompson Center sees children within about 8 to 9 months.

While increasing the number of experts who can diagnose a child is one way to speed up the process, the training and resources needed by clinicians are continually outpaced by the demand for evaluations.

“Training highly qualified diagnosticians takes years of advanced education,” said principal investigator and Thompson Center Executive Director Dr. Stephen Kanne, who is also one of a handful of independent trainers on the ADOS-2, the gold-standard diagnostic measure for autism. “Keeping up with the need for services is a national issue in the field.”

Another solution? Use highly effective screening tools that funnel children who are most likely to have an autism diagnosis into a high-risk clinic, and those at low risk of having an autism diagnosis into clinics for more appropriate services, such as for learning disabilities or other disorders, with significantly shorter waits.

A new app called Cognoa, which allows parents to upload video of their children responding to predetermined prompts as well as completing a questionnaire, may fit the bill.

In a study conducted last fall, the Thompson Center led a study with two other diagnostic centers that tested Cognoa’s ability to accurately screen for children with a high-risk of autism through a complex algorithm, which produces an indication of the likelihood that the child has autism. Children ages 18 months to 6 years were offered a chance to participate in the study after first having completed the Cognoa screening process.

The app sends the uploaded video to a group of technicians trained to look for behaviors and symptoms of autism and assign each a numerical code. This data is combined with the parent’s responses to the questionnaire, producing an indication of a child’s likelihood of having autism.

With clinicians unaware of the Cognoa score from the screening, all 230 children were then given the ADOS-2 by a standard multidisciplinary diagnostic team.

The results were promising: Cognoa got the diagnosis right 71 percent of the time.

In the next phase of the study, the research team will test Cognoa with an updated version of the algorithm that the developer hopes will improve on the app’s accuracy.

“Through the wonders of machine learning and mathematics, every time more data is computed, the algorithm can adjust in response,” said Dr. Kanne. “This technology shows great promise as a litmus test for children with a question of autism and potentially reducing the wait for an expert diagnosis.”