Children with intellectual and developmental disabilities like autism are at increased risk of showing problem behaviors like self-injury, aggression and property destruction — behaviors that expose them to being hurt, removed from the classroom or hospitalized.
Supported by a four-year $1.1 million National Science Foundation Smart and Connected Health grant, Vanderbilt engineering and pediatric researchers are working with partners from North Carolina State University to study wearable technology to detect impending problem behavior in children with intellectual and developmental disabilities.
The goal of the study is to determine whether the system — which includes custom sensors worn on the wrist like a smartwatch; a shirt designed with the Vanderbilt University Department of Theatre, with sensors inside that can track body movements; a custom-made app; and expert human input — can predict the precursors of problem behavior, or subtle signs that behavior is about to occur, during sessions of applied behavioral analysis.
The study is taking place at Vanderbilt and within community behavioral therapy clinics throughout Nashville. It will involve 20 participants and their caregivers in two to three sessions. Vanderbilt therapists and engineers work with parents and behavioral therapists in those clinical spaces to collect data from children wearing the devices while they complete behavioral assessments.
The system, which also uses cameras and microphones, collects data like how fast the child’s heart is beating, how much he or she is perspiring, and how different body parts are moving, among other indicators. A computer-based system that uses artificial intelligence and machine learning models then takes all of the data and compiles it to predict if a precursor to a problem behavior is about to occur.
The goal of the wearable technology is to make behavioral assessment safer and more accurate by predicting the precursors of problem behaviors and using that information to adjust therapeutic approaches.
“Much in the same way your Apple watch can tell you, ‘Hey, you haven’t walked in a while. You should get up and take some steps,’ this is a system that might say, ‘Your patient is starting to show some signs of getting upset. They’re not super upset yet, but it might be coming. Perhaps you can use a different strategy,’” said Amy Weitlauf, PhD, associate professor of Pediatrics at Vanderbilt University Medical Center and a Vanderbilt Kennedy Center researcher.
The study is using psychologists, behavior analysts, engineers and community therapists to understand if the process can be better, safer and more accurate.
“Ultimately, this work is not predicting the unsafe or problem behavior itself, but instead recognizing the signs that a behavior is about to happen so we can prevent it. If a parent tells us that right before a child hits someone, they often shake their head from side to side, our system might then notice that before they shake their head, their heart rate spikes, and they stiffen their arms. A human therapist might not notice the arms and definitely could not notice the heartbeat. So, the parent, system and therapist work together to identify signs that something unsafe is about to happen.
“We’re trying to collaboratively use the input of people with autism and other disabilities to create something that meets a community need, getting input all along the way, instead of designing and developing something and then asking people after we’re already done,” said Weitlauf, who is co-principal investigator with James Dieffenderfer, PhD, assistant research professor in electrical and computer engineering at North Carolina State University. Nilanjan Sarkar, PhD, David K. Wilson Professor of Engineering and vice dean for the School of Engineering at Vanderbilt, is the principal investigator of the study.
“This is a truly interdisciplinary research project that involves cutting age AI and wearable computing research along with state-of-the-art behavioral intervention techniques to address a very challenging and impactful problem,” Sarkar said. “I personally feel very gratified when I see an abstract research concept translated into real-life applications of impact — this project is a perfect example of that.”
The wearable technology was developed at North Carolina State University. Sarkar and graduate students in his laboratory have developed several components of the system including the method for how the data is analyzed. A textiles expert in Vanderbilt’s theater department helped the group select the fabric for the shirt, Weitlauf said.
The investigators are working closely with John Staubitz, MEd, a board-certified behavior analyst and director of Behavior Analysis in Education with Vanderbilt Kennedy Center’s TRIAD (Treatment and Research Institute for Autism Spectrum Disorders).
“It’s a wonderful collaborative project,” Weitlauf said.
