A Machine Learning Based WSN System for Autism Activity Recognition

Abstract

Autistic children often develop abnormal habits and in some cases they could be unsafe or even dangerous to themselves and their family members. Because of their limited speech ability, their inexperienced parents may underestimate their physical abilities compared to their intellectual level and may not realize that they could easily hurt themselves. Therefore, the need for an automatic alert system for autistic child parent assistance is great, and it will enhance the life experience for both the autistic child and the family. In this paper, we present a machine learning based electronic system for autism activity recognition using wireless sensor networks (WSNs). The system accurately detects autistic child gesture and motion. The system is named Autistic child Sensor and Assistant System (ACSA), and is comprised of three main components: the ACSA Wearable sensor device, the companion ACSA Parent Application and the machine learning algorithms developed for autistic movement event detection and processing. The paper describes the system concepts, its components and details of its architecture and operation. Individuals and families with Autism Spectrum Disorder children can utilize this system as alarming devices that assist them to protect their autistic child regardless of the environment. The proposed system is expected to enhance the life experience for all aides, the autistic child, the parents, and the autistic child whole family.