At six months of age, however, they could predict the high risk girls with about 80 percent accuracy, though they couldn't do the same for boys that young.
"It seems perhaps they are on a slightly different developmental trajectory," Bosl said.
The study is published online Feb. 22 in BMC Medicine.
EEGs measure brain electrical activity through electrodes attached to the scalp. The technology has been around for awhile -- developed in the late 1920s, it has been used for more than 60 years to detect seizures in epileptics.
But it's the newer, artificial intelligence technology and sophisticated computer algorithms that enabled the researchers to look more deeply into what the EEGs show, Bosl said.
"Artificial intelligence gives us the ability to find patterns we might not find with our own eyes," Bosl said. "One of the difficulties with a disorder like autism is that it's very heterogenous. A very high-functioning person with autism might not be so different from a so-called 'normal' person who is quirky. Defining the differences may be somewhat subtle."
EEGs are also relatively inexpensive, painless and safe, Bosl said. And unlike MRIs, they require no sedation, so testing could be put to widespread use, he said.
"My hope is we would have a simple way of measuring brain activity in every child and see the patterns emerging that might track autism characteristics," Bosl said. "That would be tremendously useful. We know early intervention is extremely important. Right now, for a lot of children, that means 3 years old. What we don't know yet is if you can intervene at 9 or 12 months and how effective that could be."
Dr. Joshua Ewen, a neurologist and director of the clinical neurophysiology laboratory at the Kennedy Krieger Institute in Baltimore, said the study is well done and looks promising, but needs to be replicated.
Also, the study predicted who was at high risk of
All rights reserved