Tech hopes to develop early warning tools, and treatments, for autism

Researchers at Georgia Tech hope to create an inexpensive, computerized early warning system for young children who have autism.

The socially-isolating affliction is frightening for parents, but it can be treated. And that treatment can be more effective with early intervention, project members said.

Children, on average, are not screened by an autism expert until age 4, but the American Academy of Pediatrics recommends that all infants be screened at 18 months of age.

"The problem is," said Gregory Abowd, a computer professor at Tech and one of the team leaders, "we don't have any way to effectively do that across the entire population."

Tech is leading a team of universities that won a $10 million grant from the National Science Foundation to develop computerized tools that could be widely disseminated so regular people with minimal training could assess toddlers for autism.

"I'm fairly confident that there's a lot you can do to complement what the professionals can do," said Abowd, a professor in Tech's School of Interactive Computing.

Within a couple of years, he hopes to roll out a prototype system. And by the end of the five-year grant, he thinks they'll have a sophisticated yet simple tool that can be mass-produced so pediatricians and even daycare operators can screen children for autism.

Autism produces behaviors in children -- such as avoiding eye contact in certain situations --  that an experienced clinician can use for a diagnosis.

The system would substitute technology for experience. It would involve a smart video camera that assesses the facial reactions and eye movements of a child who is responding to instructions from an adult. And it would include a watch-like device that monitors the child's heart rate and skin electrical conductivity -- involuntary physical responses that can help establish emotional states, such as fear, that put the voluntary facial movements into context.

The computers would take all this information and boil it down into a quantifiable result for the operator, explained James Rehg, another Tech computing professor and Abowd's partner on the project.

"What we're trying to do is simply make the collection of relevant data easy to do," he said.

The computers would operate in conjunction with a four-minute protocol developed by Opal Ousley, a research psychologist at Emory University's autism center.

She developed it as a standardized process for clinicians to diagnose autism. The expert judges the child's reaction to stimuli, such as a thrown ball, a held book and the calling of the child's name.

Autistic children react differently, she said. Their face may not turn when their name is called and their eyes may not follow the ball or the book. They lack the "social reciprocity" of unaffected children, Ousley said.

She'll recruit children at Emory's autism center to work with the Tech computer scientists, and she's hopeful they can produce something that can be widely distributed.

"I don't think it's ever going to replace experts wholly," she said, "but it can help."

Another clinician who is involved with the project said he hopes the computer people can help to develop treatment tools.

Nathan Call, the director of the Behavior Treatment Clinics at the Marcus Autism Center in Atlanta, is an expert at treating children with the disorder. He said there are no standard procedures for treating autism. Instead, clinicians rely on their training and expertise to match behavioral training to a child's progress.

By working with the computer experts, Call may be able to develop standardized processes and computerized tools that would allow parents to fill in for clinicians and their expert judgment.

Call said early treatment is crucial and that the early warning system would be a useful weapon in the fight against autism.

Children blossom linguistically between ages 3 and 5. Though it's unclear why -- whether because of missed interaction with peers who are learning to talk or failure to capitalize on a developing brain -- autistic children who are treated later progress less well than those who get early intervention, he said.

"There actually is substantial research that shows that the earlier the intervention starts, the greater the impact in changing the developmental trajectory of the kid," Call said. "They're just missing out on what's happening at that age and those are years that they're not going to get back."