Wearable-Sensors-Based Platform for Gesture Recognition of Autism Spectrum Disorder Children Using Machine Learning...

Area of Science:

• Biomedical Engineering
• Computer Science
• Developmental Psychology

Background:

• Autistic individuals often face communication challenges, including difficulties with gestures.
• Understanding autistic children's unique gestures is crucial for effective interaction.
• Existing methods for gesture recognition may not adequately address the nuances of autism spectrum disorder (ASD).

Purpose of the Study:

• To develop and evaluate a wearable-sensor-based Internet of Things (IoT) platform for recognizing gestures in children with ASD.
• To enhance communication and understanding between autistic children and their caregivers or others.
• To explore the efficacy of various classification techniques for autistic gesture recognition.

Main Methods:

• A dataset of 24 distinct gestures was collected from 10 autistic children using wearable sensors.
• Sensor data was transmitted via Bluetooth to a server for acquisition and classification.
• Time- and frequency-domain features were extracted and analyzed using k-nearest neighbor (KNN), decision tree, neural network, and random forest models.

Main Results:

• The proposed platform demonstrated high accuracy in classifying autistic gestures, achieving approximately 91% with multiple classifiers.
• Both dataset cross-validation and leave-one-person-out cross-validation methods confirmed the system's robust performance.
• The study successfully defined, monitored, and classified gestures specific to individuals with ASD.

Conclusions:

• Wearable sensor technology offers a promising approach for recognizing and interpreting gestures in children with ASD.
• The developed IoT platform can significantly aid in understanding and supporting the communication needs of autistic individuals.
• This technology has the potential to improve social interaction and daily life for people with autism spectrum disorder.