Autism Detection based on Deep Learning

Abstract

Autism Spectrum Disorder (ASD) is a complex developmental condition that affects communication and behavior, with prevalence rates increasing significantly in recent years [1]. According to recent research, early detection remains a challenge but is essential for effective intervention. This study leverages deep learning, specifically the ResNet 34 model, to analyze facial features in children, facilitating early detection of ASD. Using cross-validation to ensure robust model performance, the approach achieved an accuracy rate of 87% with ResNet 34 and 86% with cross-validation. This study contributes to the field by offering a non-invasive diagnostic aid that can help healthcare providers recognize ASD traits through facial analysis. The findings highlight the potential of deep learning in advancing ASD detection, with future work aimed at expanding the dataset and improving model precision.