Special needs

An accurate test for predicting autism in babies could be on the horizon

A promising new diagnostic tool could make it possible to screen babies for ASD, so they can receive offer earlier interventions.

An accurate test for predicting autism in babies could be on the horizon

Photo: iStockphoto

For now, the most reliable way to diagnose autism spectrum disorder (ASD) is by observing behaviours that don’t clearly emerge until around age two. These might include: repetitive hand flapping and rocking; poor eye contact or failure to engage with other kids; and sensitivities to bright lights or noisy environments. But the search is on to find an objective diagnostic method, based on biological clues.

Many practitioners currently use a screening questionnaire called M-CHAT, done at 16 to 30 months. There are 20 questions on the form, such as: Does your child walk? If you point at something across the room, does your child look at it? Does your child play pretend or make-believe? Parents can complete this questionnaire at home then bring it to their child’s practitioner to discuss whether further assessment is needed.

One drawback of this method is that the symptoms of ASD can look different in each child, and since autism is a disorder on a spectrum, those symptoms may be subtle or pronounced. That means everything comes down to the subjective conclusions of the diagnostic specialist, usually a child psychologist or specially trained physician.

While a diagnosis is typically made at the age of two years, sometimes an individual with a less classic presentation of ASD might even pass under the radar into their teen or adult years without the condition being identified. This is particularly a problem for girls, whose ASD symptoms are typically less disruptive and whose social challenges are often masked, so they go undetected.

However, a new study, published in the journal Scientific Reports, suggests there is a test that can predict ASD with astonishing accuracy in a child aged three to nine months. The predictive tool is an electroencephalogram (EEG), a fairly cheap, noninvasive test measuring brain activity that some researchers think could lead to earlier interventions, and better outcomes later in life.

In the study, the EEG captured relevant abnormal brain activity in infancy, which is important for care moving forward, believes William Bosl, a professor at the University of San Francisco and Harvard Medical School and co-author of the new study.

“Relatively minor brain problems, for example those that impair recognition of facial expression at six to 12 months, can prevent normal development of social skills. You need to process facial expressions to learn to interact, to understand emotions, and ultimately learn typical social interaction. Early detection of such processing issues opens a window for clinicians to develop new interventions to strengthen impaired processes,” says Bosl.

Bosl and his colleague analyzed EEG data from 99 infants who had a sibling with autism and compared it to data from 89 who did not. The EEGs were periodically done from three months to 36 months, producing complex computer algorithms showing differences in how the brain processes and integrates information.


These algorithms predicted autism with almost 100 percent accuracy by nine months and accurately predicted severity of symptoms by that age.

But it’s one study among many trying to figure out biological markers – in this case, brainwaves –to predict autism. It’s a good first step but it’s not ready for clinical use, says Evdokia Anagnostou, an autism researcher at Holland Bloorview Kids Rehabilitation Hospital in Toronto.

“There are many different biologies underlying autism so to confirm a marker predicts autism you must look at many groups of kids. … and even when we get good biological markers there are other things we need to think of to change our practice. For example ease of using the marker and accessibility are as important as its reliability,” she says.

Anagnostou’s concern is the test may be complicated to analyze “so it’s early to say if it could be adopted widely and used in most primary care settings.”

Charles Nelson, a professor at Harvard Medical School and at Boston Children’s Hospital who co-authored the study believes EEG has potential for fairly broad application.


“The EEG is completely non-invasive and simple to use. There are no constraints on who can have it done, and it takes only a few minutes to perform. Thus, this technique could easily be built into a pediatric practice. And we might one day be able to screen all infants for autism,” says Nelson.

So far, there is no official test based on biological markers to diagnose ASD, but research continues in pursuit of this goal. The problem is that the causes of the condition are unknown, and there are many types of autism, so it’s a complex scenario.

Recently, MRI has also shown some promise as a predictive tool. But these imaging tests are very expensive and not all infants will tolerate them, says Nelson. While there has also been research into gene tests, he says, “I have little faith in genetic markers: Autism is too complex to lend itself to a simple genetic test. Plus while 15 percent of cases have a genetic cause these seem to be rare disorders… I think looking at the brain is the way to go.”

As the search for more answers continues, Anagnostou says what parents should know for now is: “It’s unlikely we will find just one biomarker. There are many types of autism, so there can be  many biologies, and there will be different biomarkers for different groups of children.”

This article was originally published on May 10, 2018

Weekly Newsletter

Keep up with your baby's development, get the latest parenting content and receive special offers from our partners

I understand that I may withdraw my consent at any time.

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.