Autism identification and learning through motor gesture patterns

Autism Spectrum Disorder (ASD) is a mental development disorder which affects the processing of information in brain causing delays in the language and reasoning development. A person's behavior can make it distinguishable if he/she has the disorder or not through three main traits i.e. social interaction, restricted/repetitive behavior or communication skills. ASD is usually diagnosed clinically on the basis of some core symptoms and tests which are expensive and can be stressful for a child. Research has shown that Autistic people can be differentiated through the difference of force in their Motor Gestures which is related to restricted/repetitive behavior of Autistic individuals. Motor disturbance is observed in autistic patients which is a key factor for us to make its identification possible through a technical early stage approach which can be very effective. A preschool level Android touch application has been developed in which the motor patterns are observed through touch sensors and inertial sensors. Ten features were obtained as kids touched the screen for the gameplay. This touch and inertial data was then analyzed by processing them through Machine Learning (ML) algorithms to classify between Autistic and Non-Autistic group on the basis of the ten features and the motor disruption was identified. This data supports the motion disruption of movements which is identified to be as a core feature of ASD (restricted/repetitive behavior) and this demonstrates that Autism is possible to be computationally assessed through an interactive and fun gameplay on a smart device. In this way we will be able to diagnose Autism through an easy to access, learning smart phone application instead of a typical, expensive clinical diagnosis.