Skip to content

New Stanford research offers hope for faster autism diagnosis

What's the first step in getting help for a child who may have autism? Discouragingly, the answer is often "A long wait."

The number of children diagnosed with autism is growing quickly — it's now 1 of every 68 American kids — and there aren't enough specialists to go around. Early treatment helps kids with autism, studies have shown, but before they can start, children who need an in-depth clinical evaluation with a developmental pediatrician may now spend as long as a year on a wait-list.

Stanford systems biologist Dennis Wall, PhD, wants to crack open that bottleneck. His research team is using a big-data approach to devise simple questionnaires that enable parents and primary-care doctors to screen children for developmental disorders using a mobile device. Recently, his team published two papers that describe important advances in that work. And a third online project that's now open for wide participation is expanding the power of the existing findings (more on that below).

"Waitlists include many kids who don't need to go through a full clinical evaluation because they have such clear symptoms of autism or of other developmental delays," Wall told me. "We'd like clinicians to be able to act on diagnoses on the spot and say 'It's obvious that this child should be diagnosed with x or y,' or else 'This is an especially challenging case' and triage a child quickly down a more traditional path of clinical testing."

Wall's new studies use mathematical algorithms to zero in on small sets of screening questions that best identify autism and other developmental disorders. This differs from how screening tools were developed in the past, when experts used their clinical experience and intuition to come up with long lists of diagnostic questions that were then validated with patients. Wall's team is using computers to pare down the old, long lists to only those questions that give the best indication of children's likely diagnoses. The resulting brief questionnaires can be scored by computer, too.

One of his new studies, published Feb. 12 in the Journal of Autism and Developmental Disorders, describes a set of seven questions called the Mobile Autism Risk Assessment. It asks about a child's ability to use language, engage in imaginative play and interact with other people. It takes five minutes for a parent to complete, and works as well or better than alternate autism screening tools.

The other paper, published Feb. 9 in one of Nature's journals, Translational Psychiatry, addresses how to distinguish children with autism from those with attention-deficit hyperactivity disorder, which has overlapping symptoms. Starting with a 65-item screening tool called the Social Responsiveness Scale, Wall's team pinpointed five questions that can sort kids into one category or the other.

In addition to quickly identifying children who clearly have a particular developmental disorder, Wall says his new tools will make "watchful waiting" easier in cases where it's warranted. Especially before age three, it's hard to tell whether or not a child's delay in meeting developmental milestones represents a problem or a variation of normal learning.

"If at a 12-month well-baby check there's some cause for concern, many pediatricians now just say, 'Come back in six months,'" Wall said. In the future, these doctors could ask parents to fill out a mobile questionnaire several times during the wait, giving data about how a child's development progresses over time. "Then doctors can be more quantitative about the decision to provide a referral or diagnosis," he said. Even physicians with little experience of developmental disorders would be able to compare a child to large groups of other similar patients.

Speaking of large groups of patients, Wall's team is now running an online research project to collect data on thousands of children with and without developmental disorders. Any parent of a child aged 2 to 17 can participate by filling out an online survey about their child's social behavior and interactions with peers. The survey takes about 10-15 minutes and is a great way to get a window into Wall's work.

Wall doesn't think that making these diagnoses should ever be completely automated, but hopes his work will help autism specialists use their limited time better. "There will always be a need for human interaction in the diagnostic process, but it is a good goal to try to create increasingly more mobile, family-friendly tools to send children down a diagnostic path as efficiently as possible," he said.

Previously: Using Kinect cameras to automate autism diagnosis, Home videos could help diagnose autism, says new Stanford study and Using Google Glass to help individuals with autism better understand social cues
Photo by Travis Wise

Popular posts