Implementation of Visual Tracking Model for Assessing Attentive Behavior in Children with Autism Spectrum Disorder Through Eye-Movement

Eye movements provide insights into attentiveness and focus, which are crucial for determining fixation. This study develops a model for assessing attentiveness in children with autism spectrum disorder (ASD) using an eye-tracking system. The study aims to aid trained specialists in assessing attention deficiency to apply necessary mental health care to children. An eye tracker was developed to plot the gaze of thirty-four children aged 3–12 using the dlib Gaze Tracking library. Data augmentation techniques such as rotation, flipping, and filling were applied to enhance the training dataset. The model was implemented in an eye-tracker system as a proof-of-concept for the viability of early screening and intervention solution in assessing attentive behavior in children with ASD. The implemented Convolutional Neural Networks (CNN) with the K-means model resulted in 96% accuracy and 94% recall, indicating reliability in predicting attentiveness.