Discover SPARK

Predicting Autism in Babies Before the Signs Appear

Marina Sarris

Date Revised: March 7, 2024

After her son was diagnosed with autism, Misty Brooks learned that autism could run in families. So when her youngest child, Camille, began to develop signs of the condition, she knew what to do. “Services started pretty early for Camille, which honestly helped her a lot,” says Brooks, a member of the SPARK autism study.

What if we could begin interventions in children like Camille before they develop the behaviors that signal autism, which often do not appear until 18 to 24 months old?

That’s the ultimate goal of an almost 20-year-old autism study called the Infant Brain Imaging Study, or IBIS. Researchers are trying to develop a way to identify autistic babies before their first birthdays. They would use that prediction to create autism therapies that can begin months to years earlier than typically occurs now, says neuroscientist Jessica Girault, Ph.D., an IBIS investigator and an assistant psychiatry professor at the University of North Carolina’s medical school.

In the United States, the average age of autism diagnosis is 4, although doctors say that autism can be reliably detected by 2. Early intervention for developmental delays can begin before an autism diagnosis. But for many children, an autism diagnosis is their gateway to the early intervention and school services.

IBIS is targeting the baby brothers and sisters of autistic children because their chance of also having autism is estimated to range from 6 to 25 percent.1 That’s much higher than the autism rate in the general population, which is 3 percent.

One brain scan at a time, IBIS has been charting subtle changes in the brains of babies who later develop autism. “Much of what we know about brain development in the first years of life, before and alongside the emergence of autism symptoms, comes from this study,” Girault says.

Watching a Baby’s Brain Develop

During the study, IBIS researchers perform developmental tests and brain scans on children. They use Magnetic Resonance Imaging (MRI), which uses magnets rather than radiation, to take pictures of a baby’s brain at 6 months, 12 months, and 24 months of age. During those visits, researchers also conduct an electroencephalogram (EEG), a test that uses electrodes on the scalp to monitor electrical activity.

Researchers monitor the children to see how they develop over time and who eventually receives an autism diagnosis.

Researchers have found that the brains of autistic children develop differently during infancy and toddlerhood. “We found that both the structure and the function of the brain develop differently in autism, compared to typically developing children, and that these differences are observed in areas that are important for a variety of cognitive tasks,” Girault says.

For example, researchers noticed that the brain’s cerebral cortex, the outermost layer that is involved in thinking, motor, and sensory abilities, grows faster in autistic children. These children also tend to have more cerebral spinal fluid around the brain. This fluid helps clear waste products from the brain, Girault explains.

Also, the brain’s white matter ─ which acts as a highway carrying information between different areas of the brain ─ develops differently in autistic babies. At first, these highways seem more mature in infancy, but over time their development slows down.

These changes are small. In fact, they are unlikely to be seen by a radiologist reading a standard brain MRI scan of these children.

IBIS researchers are finding these changes by using advanced scanning techniques, along with computer algorithms that detect changes that are not visible to the naked eye, Girault explains.2,3

Enrolling Families in a New Phase of Autism Research

IBIS researchers have begun a new phase, which will see whether the methods developed so far can be used to predict autism in a new group of children. If this phase gets similar results, it will provide more evidence that these methods work and can be used to develop very early interventions for autism.

“We’ve laid the foundation for what we think is happening in brain development. Now, can we replicate that and build a broader picture of brain development?” Girault asks.

Researchers are enrolling 250 babies no older than 6 months, each of whom has an older sibling on the spectrum. Someone who is pregnant can enroll before their baby is born. Families will travel to the closest IBIS site, at University of North Carolina, Children’s Hospital of Philadelphia, Washington University in St. Louis, University of Washington, or University of Minnesota. They will receive compensation, reimbursement of travel expenses, and reports about their child’s development.

Several dozen families have enrolled in IBIS through SPARK Research Match, which connects SPARK participants to autism researchers in the United States and abroad.

Enter the “Baby Whisperers”

At the IBIS site, an MRI technician will scan the babies’ brains while they sleep, without any sedation. To do this, parents first take the babies to a small room where they can rock them to sleep. Once asleep, the babies move to the scanner room. A technician puts ear protection on the baby and climbs into the scanner tube with them.

“The scan techs, who we lovingly refer to as the baby whisperers, go into the scanners with the infant, keeping their hand on them,” Girault says. The techs make sure the babies are sleeping comfortably. If the baby wakes up, the tech can stop the scan.

Babies who do not fall asleep naturally will not be scanned, but among those that do, from 80 to 90 percent of the scans result in pictures that are usable, she says.

The Future of Early Detection of Autism

The MRI is a useful tool to predict autism in the baby siblings of autistic children, Girault says. But that does not mean that MRIs, which are expensive and time consuming, will replace autism screening surveys in doctors’ offices. “We don’t necessarily think that MRI is going to be a catch-all screening tool that will be used with the general population,” she says.

“We could imagine a future down the road where there are screeners that are easier to use and less costly, like maybe eye tracking,” she says. For example, researchers have developed a device that tracks youngsters’ eye movements as they watch a video of children interacting with each other. The device can help predict autism in children as young as 16 months old.4,5

“Maybe MRIs could be a useful, second-level screener in that context,” she says. But for now it’s a way to speed up identification of babies so they can enter a study of effective early interventions for autism as early as 12 months, she says.

The goal of this intervention is to reduce the effects of some aspects of autism, such as communication, learning, and social delays or difficulties.

“We want to intervene before these behaviors arise because once they’ve arisen, it might be harder to modify them,” Girault says. “We want to provide children with the best possible set of tools to guide their learning across early development, so that they can get the most out of learning when their brain is growing and ready to receive that information.”

Misty Brooks, the SPARK parent, agrees that early intervention makes a difference. Her children, who are now 6 and 7, each began therapy when they were 2 years old, months before their autism diagnoses. “Early intervention changed the outlook of my children’s lives,” she says. “If they didn’t have that intensive early intervention in the home, I don’t think they would be where they are today.”

To learn more about sibling research, watch the SPARK webinar, “Infant Visual Brain Development in Autism,” and see a list of IBIS papers on the publication page.

Interested in joining SPARK? Here’s what you should know.

Photo credit: iStock


  1. Palmer N. et al. JAMA Pediatr. 171, 1107-1112 (2017) PubMed
  2. Hazlett H.C. et al. Nature 542, 348-351 (2017) PubMed
  3. Emerson R.W. et al. Sci. Transl. Med. 9, eaag2882 (2017) PubMed
  4. Jones W. et al. JAMA 330, 854-865 (2023) PubMed
  5. Jones W. et al. JAMA Netw. Open 6, e2330145 (2023) PubMed