Autism spectrum disorder affects a significant number of children both worldwide and in the United States.
While there is no cure for the condition, early detection can make a positive difference in a child’s life.
In a new study, researchers say they have found a way to accurately predict autism onset in infants who are as young as 6 months.
Detecting autism as early as possible has a positive bearing on a child’s health outcomes and overall well-being.
But so far, apart from knowing about a few risk factors that increase the odds of developing ASD (such as being a boy or having a sibling diagnosed with autism), there are no means of detecting ASD in children before they start to display symptoms.
Generally, children start showing signs of autism by the age of 12 or 18 months. Based on their communication and behavioral symptoms, children are typically diagnosed when they are around 2 years old.
But new research, published in the journal Science Translational Medicine, may have uncovered a way to predict autism earlier than this.
Using a neuroimaging technique called functional connectivity magnetic resonance imaging (fcMRI) together with machine learning software, researchers from the University of North Carolina (UNC) in Chapel Hill were able to predict which 6-month-old high-risk infants would develop ASD by the age of 2.
Building on previous research
A study published in the journal Nature earlier this year showed that brain changes at 6 and 12 months can help to predict autism in high-risk children.
The senior author of the new research explained the impact of their study’s contribution:
“The Nature paper focused on measuring anatomy at two time points (6 and 12 months), but this new paper focused on how brain regions are synchronized with each other at one time point (6 months) to predict at an even younger age which babies would develop autism as toddlers,” said Dr. Joseph Piven in a statement. “The more we understand about the brain before symptoms appear, the better prepared we will be to help children and their families.”
Piven is a professor of psychiatry at the UNC School of Medicine as well as director of the Carolina Institute for Developmental Disabilities.
For the new study, Piven and his colleagues placed 59 infants that had a high risk of developing autism inside an MRI machine. The babies were 6 months old and they were sleeping naturally at the time of the testing.
Using fcMRI, the researchers analyzed how the different regions of the brain connect and work together to perform cognitive tasks, as well as during rest.
Piven and his team used a machine learning classifier, which is a computer program that can group the differences found in neuroimaging results into two categories: autism and non-autism.
The classifier used the links it learned between synchronized brain regions and later behavior to predict autism diagnoses.
At 2 years old, 11 of the 59 infants received an ASD diagnosis. Using machine learning, the researchers managed to correctly identify nine out of 11 of these infants (82 percent).
Additionally, the technology accurately identified all of the children that did not go on to develop autism.
Robert Emerson, Ph.D., a former UNC postdoctoral fellow and first author of the study, explains the results.
“When the classifier determined a child had autism, it was always right. But it missed two children. They developed autism, but the computer program did not predict it correctly, according to the data we obtained at 6 months of age,” he said. “No one has done this kind of study in 6-month-olds before, and so it needs to be replicated. We hope to conduct a larger study soon with different study participants.”
Dr. Diana Bianchi, director of the National Institute of Child Health and Human Development, also weighs in on the results.
"If future studies confirm these results,” she said, “detecting brain differences may enable physicians to diagnose and treat autism earlier than they do today."
Dr. Joshua Gordon, Ph.D., director of the National Institute of Mental Health, also commented, saying, "Although the findings are early-stage, the study suggests that in the future, neuroimaging may be a useful tool to diagnose autism or help healthcare providers evaluate a child's risk of developing the disorder."
“The most exciting work is yet to come,” Emerson says, “when instead of using one piece of information to make these predictions, we use all the information together. I think that will be the future of using biological diagnostics for autism during infancy.”