Published: Thu, February 16, 2017
Research | By Elizabeth Houston

AI Predicts Autism From Infant Brain Scans


In collaboration with computer scientists at UNC and the College of Charleston, the team built an algorithm, trained it with the brain scans, and tested whether it could use these early brain changes to predict which children would later be diagnosed with autism. "It's been a continual goal of Autism Speaks and the autism community to drive the age of diagnosis to be as early as possible", said Mathew Pletcher, Autism Speaks' interim chief science officer.

By scanning the brains of babies whose siblings have autism, researchers say they have been able to make reasonably accurate forecasts about which of these high-risk infants will later develop autism themselves.

The predictive power of the team's findings may inform the development of a diagnostic tool for ASD that could be used in the first year of life, before behavioral symptoms have emerged.

One in every 100 people has autism, which affects behaviour and particularly social interaction. Usually, babies that have otherwise progressed normally will start showing subtle changes in behavior: difficulty focusing or speaking with others, or trouble pointing at objects. They scanned each child's brain-no easy feat with an infant-at 6-, 12-, and 24 months.

The brains of the autistic children had faster rates of growth on the surface of their brains from 6 months of age to 1 year, and faster overall brain size from 1 to 2 years of age.

Researchers say by identifying the brain changes early, there's the possibility of developing therapies and drugs before the brain fully forms.

Autism rates have sharply increased since the early 1990s, which helped encourage the now thoroughly debunked (although still hotly-contested) theory that vaccines can cause the disorder.

For the study, published this week in Nature, researchers conducted MRI scans on 150 children three times: at six months old, one year and two years.

Watch Holly Holm, Anderson Silva fights online, TV listings and time
He was unsure fighting would be financially lucrative enough to live in the United States and support his five children. UFC 208 is unlikely to break any box office records but there's certainly enough on the card to justify a late night.

What to Know About Recent Immigration Raids in US Cities
Castro said the roundups were part of Operation Cross Check - a series of large-scale raids that began in 2011 under Barack Obama. ICE called the reports of massive round-ups "false, risky and irresponsible", the New York Times reported.

PM Modi to address rally in Badaun
It's the first chance for Modi to see if abolishing 86% of the cash circulating to crack down on corruption is hurting his appeal. On Saturday, Gandhi further said that Modi had not fulfilled the promises he and his party made during the elections in 2014.

The study was funded by the National Institutes of Health and led by a team of researchers at the University of North Carolina-Chapel Hill.

"We now have this finding in these high familial risk infants that we can predict 8 out of 10 that we think will get autism", says Piven, adding that behaviour-based predictions do no better than 50-50 at that age.

Research could then begin to examine interventions on children during a period before the syndrome is present and when the brain is most malleable. The researchers note that the greater the brain overgrowth, the more severe a child's autistic symptoms tended to be.

Schultz, Ph.D., is the director of the Center for Autism Research at Children's Hospital of Philadelphia.

The study also has implications for developing new autism treatments, said Schultz, a pediatric neuropsychologist.

However, by considering other factors as well including additional brain measurements and the child's sex, the researchers used a statistical approach known as machine learning to assess with near ideal accuracy who would develop autism.

'Putting this into the larger context of neuroscience research and treatment, there is now a big push within the field of neurodegenerative diseases to be able to detect the biomarkers of these conditions before patients are diagnosed, at a time when preventive efforts are possible, ' Piven said. "Using brain imaging, we were able to pinpoint areas of the brain where atypical development contributes to autism".

'In Parkinson's for instance, we know that once a person is diagnosed, they've already lost a substantial portion of the dopamine receptors in their brain, making treatment less effective'. "The fact that they're not consistent suggests that some of the expansion in surface area may actually not be relevant to the detection of autism", he says. "We haven't had a way to detect the biomarkers of autism before the condition sets in and symptoms develop", he said. Other key collaborators are McGill University, the University of Alberta, the College of Charleston, and New York University.

Like this: