THIS ARTICLE IS MORE THAN FIVE YEARS OLD
This article is more than five years old. Autism research — and science in general — is constantly evolving, so older articles may contain information or theories that have been reevaluated since their original publication date.
Researchers have adapted a motion-sensing video game controller to detect repetitive movements that are characteristic of autism, according to a preliminary study published in the 2012 Conference Proceedings – IEEE Engineering Medical Biological Society1.
Repetitive and restricted behaviors are one of the core symptoms of autism. This category includes both physical behaviors, such as hand flapping or body rocking, and more general traits, such as an insistence on sameness or obsessive interests.
Despite their status as a core symptom, physical repetitive behaviors are not always harmful and may in fact be soothing to an individual with autism. For example, children with the disorder often ramp up these behaviors when confronted by a new situation, or when transitioning from one task to the next, suggesting that the behaviors signal anxiety or act as a coping mechanism.
To study these movements, researchers rely on questionnaires filled out by caregivers or examine videos of the child. More objective and quantitative approaches include accelerometers, which detect movement. But children may object to wearing these devices, the researchers who led the new study say.
The researchers instead developed a system that can detect repetitive movements from anywhere in a room without requiring the child to wear a specific device. Their goal is to use the system to assess the effectiveness of a behavioral intervention.
The system uses a Microsoft Kinect video game controller, which is outfitted with a camera and a depth sensor and can detect movement in three dimensions. The device relays information about the child’s movements to a software application for analysis.
The Kinect sensor recognizes the distance between itself and a number of preprogrammed points on the child, such as his hands, wrists and head. Changes in this distance in any one dimension create patterns that reflect the child’s movements. For example, repeated spikes and dips in one dimension indicate hand flapping.
The researchers then trained the software to recognize these movement patterns. The trained software samples movement across five sequential frames taken by the camera. If it identifies a pattern that resembles hand flapping, it begins timing the behavior.
Researchers tested the accuracy of the sensor by simulating hand flapping themselves and found that it accurately detects the behavior. However, the software did report hand flapping in a few instances when the repetitive behavior was not present.
The researchers plan to fine-tune the device to eliminate these errors. They also intend to develop a measure of body rocking, another repetitive behavior common in autism. Ultimately, their aim is to use the device to score children’s reactions to a robot, which they use as a therapeutic tool for the disorder.
1: Goncalves N. et al. Conf. Proc. IEEE Eng. Med. Biol. Soc. 2012, 1598-1601 (2012) PubMed