Duke researchers hope new app can help more quickly diagnose autism

DURHAM, N.C. (WTVD) -- A team of Duke doctors, computer scientists and engineers are working on an app that could help more quickly diagnose autism in young children.

"If you think about autism, it is a behavioral diagnosis. It's based on observations of behavior. So how do you develop a tool that can precisely measure behavior in an accurate way that could pick up on really subtle signs in many cases early in life? And that was our challenge," explained Dr. Geraldine Dawson, a Professor in Psychiatry and Behavioral Sciences at Duke University School of Medicine.

According to the CDC, just under 3% (1 in 36) of children have been identified with autism spectrum disorder.

Dawson said one key tool currently used to diagnose kids is a parent questionnaire, often administered during an 18 or 24-month check-up. However, Dawson said the questionnaire can miss signs and has not proven as effective in girls or with families of ethnic minorities.

She said she believes a new app, created by the research team with funding from Duke and NIH, could help address these shortcomings. It includes a series of prompts and videos for a young child to view and respond to, with reactions captured by a camera.

"Those videos are then automatically coded using a technique called computer vision. So this is where the computer can precisely measure all the movements that the child is using. How is their face responding? What are they looking at? Are they making any other body movements? And it turns out that those signs are ones that can help us detect an early diagnosis of autism," said Dawson.

While the app itself cannot diagnose autism, its data points can be utilized to help doctors more quickly come to a diagnosis.

"We have not had those objective quantitative measures in autism before. And that's what we feel is needed. So we would combine what the parent is telling us about what they're seeing and also what the pediatrician is saying with this quantitative and objective approach, which we think together will help us make sure that we're not missing any kids," said Dawson.

It's a tool that Dr. Danai Fannin said she believes could have helped her two children when they were younger.

"There were things they would do, like kicking their legs. Definitely, their eye gaze was different," said Fannin.

A speech pathologist who works at NC Central and Duke, Fannin recalled both children experienced processing and development delays.

"My son wasn't talking until he was 18 months old. And usually, a child should have a word before then. So that's when he first started talking. But we just thought he was the basic late talker. With my daughter, she was not talking. And when we would call her name, she wouldn't respond. There were some things, like gestures she didn't do (at 14 months)," Fannin recalled.

There were other health issues, though Fannin said they were addressed individually.

"Parents end up going to the doctor for these individual things. We prioritize asthma. They were diagnosed with asthma at six months and four months. That was the priority. The feeding issues, those were the priorities. So we kind of put thoughts of developmental disorder on the back burner," Fannin said.

Both Fannin and Dawson are hopeful additional information could allow children to receive early intervention.

"It can make the difference between whether they learn to speak or don't, whether they can go to a general classroom or not. And so finding kids early in life is important," said Dawson.

"If you don't get early intervention, you miss out on advice from a nutritionist, occupational therapist, speech pathologist, physical therapist. There are so many things that you miss out on in the birth to three years (old). That's when the brain is growing rapidly, and they're learning a lot," said Fannin.

The app is in a pivotal trial, which can take another couple years to complete. To this point, it has proven effective across all demographics, addressing one existing shortcoming.

"I think it'll be really helpful to help close some of those disparities in early identification of underrepresented populations," said Fannin.