EyeXplain Autism: Interactive System for Eye Tracking Data Analysis and Deep Neural Network Interpretation for Autism Spectrum Disorder Diagnosis

Ryan Anthony Jalova de Belen, Tomasz Bednarz, Arcot Sowmya

Abstract: Over the past decade, Deep Neural Networks (DNN) applied to eye tracking data have seen tremendous progress in their ability to perform Autism Spectrum Disorder (ASD) diagnosis. Despite their promising accuracy, DNNs are often seen as ’black boxes’ by physicians unfamiliar with the technology. In this paper, we present EyeXplain Autism, an interactive system that enables physicians to analyse eye tracking data, perform automated diagnosis and interpret DNN predictions. Here we discuss the design, development and sample scenario to illustrate the potential of our system to aid in ASD diagnosis. Unlike existing eye tracking software, our system combines traditional eye tracking visualisation and analysis tools with a data-driven knowledge to enhance medical decision-making for physicians.