چکيده
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterized by persistent difficulties in social communication, restricted interests, and repetitive behaviors, typically presenting in early childhood. Early diagnosis is critical to improving developmental outcomes, yet traditional diagnostic methods—such as observational assessments and behavioral interviews are often subjective, time-consuming, and dependent on clinical expertise, which can result in delayed intervention. As ASD prevalence continues to rise, now affecting approximately 1 in 36 children, there is an urgent need for more efficient, accurate, and accessible diagnostic tools. Artificial Intelligence (AI), particularly Machine Learning (ML), offers a promising approach by analyzing large and complex datasets to identify patterns indicative of ASD that may be missed by human observation. This study investigates various AI-based prediction techniques, including Support Vector Machines, Convolutional Neural Networks (CNNs), BERT-based language models, and federated learning systems, to evaluate their diagnostic accuracy and practical applicability. The study also explores the integration of multimodal data such as facial images, behavioral patterns, language use, and eye-tracking to enhance model robustness and generalizability. Additionally, the study incorporates explainable AI frameworks to improve clinical trust and interpretability, ensuring broader adoption in real-world diagnostic settings. Ultimately, this work earlier detection, reduce diagnostic delays, and enable more personalized intervention strategies for children with ASD through intelligent, data-driven systems.
