Scientists closer to being able to predict autism

Scientists closer to being able to predict autism


Early diagnosis of autism can improve treatment results, and researchers may have found a way to detect the condition earlier.

In a study funded by the National Institutes of Health, researchers at the University of North Carolina at Chapel Hill and the Washington University School of Medicine in St. Louis accurately predicted the diagnosis of autism in children using a functional MRI technique on the brain. The study focused on infants who had siblings with autism, putting them at a high risk for the disorder.

A computer technology called machine learning sought differences in brain connections between children with and without autism and then used these differences to predict autism after scanning the brains of infants. The researchers looked specifically at how regions of the brain work together during different tasks and during rest, which is where the predictability of autism lies, according to the NIH.

The distinct social deficits and repetitive behaviors of autism usually show themselves at age 2, but brain-related changes can happen even earlier, according to the study. The researchers predicted the development of autism at that age with 96 percent accuracy.

In the study, published in Science Translational Medicine, researchers scanned 59 6-month-old children who were considered to be at a high risk for autism. The researchers identified nine of 11 children who would be diagnosed with autism at age two and correctly identified every child that would not be diagnosed with autism.

Researchers say the findings need to be reproduced but that the new technique may lead to a future where physicians can diagnose autism before symptoms appear, using a single brain scan.

Related Episodes