Data Mining of Intervention for Children with Autism Spectrum Disorder

Studying progress in children with autism spectrum disorder (ASD) is invaluable to therapists and medical practitioners to further the understanding of learning styles and lay a foundation for building personalised intervention programs. We use data of 283 children from an iPad based comprehensive intervention program for children with ASD. Entry profiles - based on characteristics of the children before the onset of intervention, and performance profiles - based on performance of the children on the intervention, are crucial to understanding the progress of the child. We present a novel approach toward this data by using mixed-variate restricted Boltzmann machine to discover entry and performance profiles for children with ASD. We then use these profiles to map the progress of the children. Our study is an attempt to address the dataset size and problem of mining and analysis in the field of ASD. The novelty lies in its approach to analysis and findings relevant to ASD.