Mobile App Uses Machine Learning to Screen Toddlers for Autism

A mobile app was able to distinguish toddlers diagnosed with autism spectrum disorder using machine learning, indicating that the technology could someday serve as a scalable early screening tool.

In a study published in JAMA Pediatrics, researchers from Duke University strategically designed movies that would allow them to evaluate a young child’s preference for looking at objects more than at people.

Research has shown that the human brain is hard-wired for social cues, and that a person’s gaze automatically focuses on social signals. In autism spectrum disorder, attention to social stimuli is reduced. Researchers have aimed to screen for autism in young children by tracking their eye movements while the view social stimuli.

However, equipment used for visual tracking is expensive and requires specially trained personnel, limiting its use outside of laboratory settings.

For the study, researchers enrolled 933 toddlers ages 16 to 38 months. The team developed several movies for toddlers to view in the app. For example, one movie shows a cheerful woman playing with a top. She appears on one side of the screen while the top she is playing with is spinning on the other side.

Researchers recorded the children’s gaze patterns with the device’s camera and measured them using computer vision and machine learning analysis. While toddlers without autism scanned the entire screen throughout the video and focused more often on the woman, the 40 toddlers later diagnosed with autism more often focused on the side of the screen with the toy.

This method is far more accessible than existing approaches: Researchers noted that the app takes less than ten minutes to administer and uses the front-facing camera to record the child’s behavior. It only requires an iPhone or iPad, making it readily available to primary care clinics and usable in home settings.

"We know that babies who have autism pay attention to the environment differently and are not paying as much attention to people," said Geraldine Dawson, PhD, director of the Duke Center for Autism and Brain Development, and co-senior author of the study.

“We can track eye gaze patterns in toddlers to assess risk for autism. This is the first time that we've been able to provide this type of assessment using only a smart phone or tablet. This study served as a proof-of-concept, and we're very encouraged."

This eye-tracking app, featuring computer vision analysis and machine learning, could serve as a viable way of identifying young children with autism spectrum disorder.

"This was the technical achievement many years in the making. It required our research team to design the movies in a specific way to elicit and measure the gaze patterns of attention using only a hand-held device,” said lead author Zhuoqing Chang, PhD, postdoctoral associate in Duke's Department of Electrical and Computer Engineering

"It's amazing how far we've come to achieve this ability to assess eye gaze without specialized equipment, using a common device many have in their pocket.”

Ongoing validation studies are currently underway, researchers noted. Additional studies with infants as young as six months are examining whether the app-based assessment could identify differences in children who are later diagnosed with autism and neurodevelopmental disorders during the first year of life.

"We hope that this technology will eventually provide greater access to autism screening, which is an essential first step to intervention. Our long-term goal is to have a well-validated, easy-to-use app that providers and caregivers can download and use, either in a regular clinic or home setting," Dawson said.

"We have additional steps to go, but this study suggests it might one day be possible."

This study builds on recent efforts to explore autism spectrum disorder using AI. A 2020 study showed that a precision medicine method enabled by AI could lead to the first biomedical screening tool for a subtype of autism.

“Today, autism is diagnosed based only on symptoms, and the reality is when a physician identifies it, it's often when early and critical brain developmental windows have passed without appropriate intervention,” said study co-first author Dr. Yuan Luo, associate professor of preventive medicine: health and biomedical informatics at the Northwestern University Feinberg School of Medicine.

“This discovery could shift that paradigm.”