Identification Of Autism In Children Using Static Facial Features And Deep Neural Networks

Abstract

This research presents an intelligent and non-invasive framework for the early identification of Autism Spectrum Disorder (ASD) in children through the analysis of static facial features using deep learning techniques. ASD is a neurodevelopmental disorder that affects communication, social interaction, and behavioural responses, and its symptoms vary significantly among individuals. Early diagnosis is essential because timely therapeutic intervention can greatly improve cognitive development, language skills, and social adaptation. However, conventional diagnostic procedures mainly depend on behavioural observation, interviews, and clinical assessments, which are often time-consuming, subjective, and inconsistent across practitioners. To address these limitations, the proposed study explores the capability of artificial intelligence to support early autism screening through automated facial image analysis.