In Europe and North America, about one in 68 children are thought to have some form of autism. Nonetheless, getting a diagnosis can be a long, expensive, and thorny process. In a push to address this problem, researchers from the University of Strathclyde in Scotland have developed an iPad game that may be able to diagnose autism.
In their study, published in Scientific Reports, the researchers explain that new scientific evidence shows that disruption to motor timing might “underpin” the disorder and provides a testable marker with which to identify it. Machine learning data analysis has an impressive accuracy rate in diagnosing children with autism by tracking key motor skills. Knowing this, the researchers collaborated with start-up tech company Harimata to apply this technology to a playable game, capable of identifying these cues purely by using gameplay data.
"Our aim was to develop a test that would be intuitive, fast, fun and engaging for the children," Anna Anzulewicz, director of research at Harimata, said in a statement. "iPad-based games seemed to be perfect, and they are embedded with powerful sensors, which allow for the precise measurement of the children's play dynamics."
Image A shows the game “Sharing” and image B shows the game “Creativity.” Image credit: Anna Anzulewicz et al/Scientific Reports/Creative Commons
They adapted two currently available games, called “Sharing” and “Creativity”, to collect data on the children's motor skills by measuring how subtly they moved their finger on the touch-sensitive screens. The “Sharing” game involved touching a fruit on the screen, slicing it into equal pieces, then putting it on four plates. “Creativity” allowed the players to trace the shape of a chosen animal and then color it in. An indication of autism could be made by recording the pressure, speed, and accuracy of the child's gameplay, and running this information through a machine learning algorithm.
"In other words, it is not social, emotional, or cognitive aspects of the gameplay that identify autism," said Dr Jonathan Delafield-Butt, a study author and senior lecturer in child development. "Rather, the key difference is in the way children with autism move their hands as they touch, swipe, and gesture with the iPad during the game."
In the study, the researchers tested out the game on 82 children, 37 of which were diagnosed with autism. The study said: “In sum, we show that children with autism can be identified with up to 93% accuracy by computational analysis of their motor pattern in iPad gameplay.”
"This is potentially a major breakthrough for early identification of autism," Dr Delafield-Butt added. "This new 'serious game' assessment offers a cheaper, faster, fun way of testing for autism. But more work is needed to confirm this finding, and to test for its limitations."
Head over to the Harimata website for more details on the app.