Enhanced Deep Learning Approach for Autism Spectrum Disorder Diagnosis Through Facial Recognition

Autism Spectrum Disorder (ASD) is a neurodevel-opmental condition characterized by challenges in social inter-action, communication, and behavior. Diagnosing autism early is critical, as early intervention significantly improves outcomes and supports developmental progress. However, traditional diagnostic processes are often time-consuming, costly, and reliant on subjective assessments, making accurate identification challenging. The proposed approach combines transfer learning and convolutional neural networks (CNNs) to facilitate early and efficient diagnosis of ASD. The proposed model leverages the power of efficientnet to analyze relevant features and detect autism with high accuracy. The model achieves accuracies of 96% and 90% for children and adults, respectively, outperforming existing diagnostic methods. The outcomes shows that the proposed model to enhance early detection and reduce the burden of traditional diagnostic procedures, paving the way for more accessible and reliable autism diagnosis solutions.