MindMeld: Design of a Brain-Computer Interaction Rehabilitation System for Children with Autism

In this work, we designed an EEG-based rehabilitation interaction system for children with autism (ASD), named the "MindMeld Autism Rehab System". The system consists of three parts: an Electroencephalogram (EEG) headband hardware, a visual game interface, and an EEG decoding algorithm for children with ASD. Based on the visualized motor imagery children’s game, ASD children can control cartoon characters through motor imagery using the EEG headband. During this brain-computer interaction process, a spatial-temporal graph neural network (STGNN) deep learning network was designed to diagnose and assess the collected EEG signals. MindMeld system combines applied behaviour analysis (ABA) rehabilitation cognitive strategies with engaging cognitive rehabilitation games, and simultaneously assesses and records the rehabilitation progress of brain disorders while ASD children are playing, providing a new interactive approach for the rehabilitation of children with autism.