The idea of applying computer science to the traditionally medical domain of autism spectrum disorders (ASDs) started with one provocative question: What if cameras and computers could capture and process information that even the most expert of experts can’t?
University of Minnesota researchers from a half-dozen disciplines are collaborating on a $3 million study to see if Kinect motion sensors—perhaps most widely known for their use in video games—and computers can detect subtle markers of ASDs and other disorders, such as attention deficit disorder and obsessive compulsive disorder, to help get an earlier diagnosis.
To find out, researchers are observing groups of 3- to 5-year-old children at the University’s Shirley G. Moore Laboratory School using several Kinect motion sensors, which capture interactions, behaviors, and movement patterns based on each child’s shape and clothing color.
“Let’s see if the cameras can detect something unusual, patterns which are unusual,” says lead researcher Nikolaos Papanikolopoulos, Ph.D., director of the University’s Center for Distributed Robotics.
For example, does a child come and go through a door when other children are around? Is that behavior different when he’s alone? How does a child engage with a toy placed in front of him or her? Does the child walk with a slight tilt?
Papanikolopoulos and his colleagues are encouraged by the details collected by the cameras and analyzed by computers.
“Imagine a room full of videotapes, and each videotape contains 10 kids doing things,” Papanikolopoulos says. “It’s very difficult for the human eye to observe, over periods of time, unusual patterns. We can see if video-based systems highlight any markers of interest.”
