Autism spectrum disorders—a group of related neurodevelopment conditions that impair communication and social interaction—are incredibly common. According to the Centers for Disease Control and Prevention, one in every 68 American children has been diagnosed with some form of autism. With intervention (the earlier the better), many autistic children flourish. Unfortunately, early diagnosis is the exception rather than the rule. Nationwide, the average age of diagnosis is four years old.
Now, a team of researchers at Duke University has released a free app, called Autism & Beyond, to mechanize part of the autism screening process. The app is available to anyone with an iPhone who is willing to take part in a Duke study.
Guillermo Sapiro, a professor of electrical and computer engineering at Duke and one of the app’s creators, says this tool is critical because there are far too few professional autism screeners to meet demand. In the United States, the average wait to see an autism specialist at a top hospital is one year. In other countries, the situation is even worse.
“The current mental health system doesn’t scale, and there’s no way we can scale it,” Sapiro says. “For us, empowering parents and caregivers is the way to go.”
Users of the Autism & Beyond app administer tests three times over a six-month period. The 20-minute tests largely involve having a child, between one and six years old, sit on a parent’s lap to watch short video clips. The videos are designed to elicit emotional responses—laughter, smiles, surprise. The phone camera records the child as he or she reacts. The app’s biometric software then analyzes these emotional responses. Duke researchers will use the app’s data to examine the software-generated analysis and see how well it works compared to human screening. If it's successful, it could ultimately be used by parents and schools in the United States and beyond.
Developed by a team of doctors, students, psychologists, programmers and engineers, Autism & Beyond was modeled after parts of traditional autism screenings, in which screeners use toys to elicit responses from children. The researchers initially tested the process in Duke clinics, replacing the human screeners with an iPad. When these results showed promise, the team moved to working on an app that could be used outside the clinical setting. The app was created with ResearchKit, Apple's open-source framework for app-based medical studies released in March.
A number of studies have already suggested that a computer is better than the human eye at detecting fleeting facial responses. A human might not recognize that a child is a fraction of a second slower than average to smile or laugh, but a computer will.
Sapiro cautions that the app is not meant to diagnose autism, merely to screen for certain worrying signs. In the future, a child who raised red flags on the app might then be guided towards an autism professional for further testing.
The engineer and his team eventually hope apps like this one could be a way to give families access to autism screening in places where traditional screening is hard to come by. The team is currently working with Chinese researchers to create a Mandarin app and with Argentinean experts to make a Spanish version. They’re also working with partners in South Africa to make the app available in various African countries—there are only 50 autism specialists on the entire African continent, Sapiro says. Eventually, they would like to create related apps to study childhood anxiety and temper tantrums.
Study participants will receive feedback on whether or not the tests suggest their children are at risk for autism, and they will be offered information about where to seek further help if any problems are identified.
“There is no doubt in the leading medical establishment that early intervention helps,” Sapiro says. “But a lot of people don’t know where to look for help.”