A system that uses A.I. to identify people with autism spectrum disorder (ASD) has been submitted to the U.S. Food and Drug Administration (FDA), and could become a valuable tool for primary care pediatricians in diagnosing autism. Its creators hope that it will help children with autism to be identified younger, thereby getting them the assistance they need sooner.
“Based on parent or pediatrician concern, a pediatrician would prescribe Cognoa’s parent-facing mobile application,” Dave Happel, CEO of Cognoa, told Digital Trends. “Once in the app, the parent answers a 10- to 15-minute questionnaire about their child’s behavioral patterns, then uploads two home videos of the child capturing their behavior in a natural environment. The videos are sent to a trained professional, who observes [them] and answers questions about their observations, which is fed into Cognoa’s algorithms. In addition, the pediatrician answers a questionnaire about the child’s behavior.”
Cognoa’s algorithm then combines these multiple inputs and compares them with a dataset consisting of thousands of other cases, covering different races, genders, and backgrounds. Happel hopes that this will help avoid some of the irregularities with current autism diagnosis, such as the Black and Hispanic children not being diagnosed and girls, on average, being diagnosed with autism later than boys.
Once it has highlighted predictive signs of autism, the idea is that the pediatrician could use this in combination with his or her own clinical judgement to make a diagnosis.
Clearing the backlog
“As it stands, the overwhelming majority of pediatricians refer children with suspected autism to specialists,” Happel said. “This has created a significant backlog for specialists and results in families facing a long wait for a diagnosis often taking months or even years — with the average age of diagnosis still being over four years of age. This is despite it being possible to diagnose autism in children as young as 18 months old.”
Cognoa has completed a double-blind clinical trial involving 425 participants, between 18 months and 72 months old. Kids in the study were assessed by both Cognoa’s system and a specialist clinician using the standard DSM-5 diagnosis criteria. These trial results have yet to be published, but the study ran from July 2019 through May 2020.
“We are currently submitting our digital diagnostic for FDA clearance and are confident of receiving clearance in the months ahead,” Happel said. “We received FDA breakthrough designation status in 2018, so we will be receiving a priority review of our digital diagnostic. We are aiming to roll out our device to pediatricians during the second half of 2021.”