Poster C90, Sunday, March 25, 1:00-3:00 pm, Exhibit Hall C
Tablet-based gameplay identifies movement patterns related to autism spectrum disorder
Anna Anzulewicz1,2,3, Krzysztof Sobota2,3, Jonathan Delafield-Butt2; 1Jagiellonian University in Krakow, Poland, 2University of Strathclyde, Glasgow, UK, 3Harimata, Krakow, Poland
It has been recently proposed that one of the early markers of autism spectrum disorder (ASD) are abnormalities in the development of intentional movements, which can be observed from early childhood. Here, we aimed to identify differences in kinematics between children with ASD and their typically developing (TD) peers. In two studies, we utilized tablet devices’ inertial sensors (accelerometer, gyroscope, and magnetometer) and touchscreen to record the movement dynamics made during gameplay. Thirty-seven ASD and 45 TD children (aged 3-6) participated in the first experiment, and 46 ASD children and 383 TD participated in the second one. Two experimental games were used: (1) ‘Sharing’ where the main task was to divide a piece of food and distribute it evenly among four children on the screen; and (2) ‘Creativity’ where gameplay was open, unstructured coloring of an object. Each game consisted of a two-minute training phase and five-minute test phase. Simple kinematic calculations of movements, e.g., gesture duration, acceleration, etc., were extracted. Machine learning algorithms (Random Forest, Regularized Greedy Forest) were used to classify movements. The results showed that the specific for ASD children movement patterns could be differentiated from the patterns of TD children with 93% accuracy in Exp.1 and 85% accuracy in Exp.2. The children with autism displayed larger and faster gesture kinematics and exerted higher pressure at a contact than the control group. These findings support the notion that autism spectrum disorder has a prominent motor component that can be identified from early childhood using smart device gameplay.
Topic Area: PERCEPTION & ACTION: Development & aging