AI Accurately Detects Early Signs of Autism in At-Risk Infants

Detecting autism (which now affects about 1 in 68 children around the world) in young babies is often difficult because the tell-tale signs, such as lack of eye contact and the failure to respond to one’s name, usually present themselves at a later age.

Colours represent the areas of the brain that grow faster than normal in at-risk infants between months six to 12. Photo courtesy of Piven et al. 2017.

With a new system, which uses an artificial intelligence algorithm to process brain scans of at-risk infants, however, early diagnosis might finally become a reality.

Given that diagnosing ASD (Autism Spectrum Disorder) commonly relies on behavioural evaluation, there are currently no scientifically approved interventions for infants considered to be at an increased risk.

For this reason, there is rarely anything to be done before symptoms begin to surface, although some researchers believe distinct changes in the brain start taking root as early as when the infant is still in the womb.

In the new study, recently published in the leading science journal Nature, a team of researchers, funded by the U.S. National Institutes of Health, scanned 106 high-risk infants at six, 12, and 24 months.

Once the study subjects reached two years of age, 15 of them were diagnosed with ASD. In line with past research, the brain volume of these infants was found to grow more rapidly than normal in the period between 12 and 24 months, at which time they also started to exhibit the first symptoms of autism.

Interestingly enough, however, certain changes were detected even earlier. The surface area of the cortex was found to increase much faster in the infants with ASD between six and 12 months.

While increased brain volume has been know to correlate with autism since the 1990s, the exact timing and direct correlation of the enlargement to ASD behavioural symptoms have not been confirmed.

The AI analysis of the MRI scans at six and 12 months allowed the researchers to forecast, with 81% accuracy, which high-risk infants would develop autism by age two.

“Typically, the earliest an autism diagnosis can be made is between ages 2 and 3,” said study lead author Dr. Joseph Piven. “But for babies with older autistic siblings, our imaging approach may help predict during the first year of life which babies are most likely to receive an autism diagnosis at 24 months.”

The ability to diagnose ASD early might help researchers test different intervention protocols in hopes of developing effective treatment modalities in the future.


Comment this news or article