Abstract
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterised by a wide range of behavioural and developmental symptoms. Understanding the risk factors that contribute to the onset and progression of ASD is essential for early diagnosis and intervention. In this paper, we explore advanced artificial intelligence (AI) approaches for analysing risk factors associated with ASD. Specifically, we apply machine learning techniques, including Random Forest, XGBoost, and SHAP (Shapley Additive Explanations) to determine the most influential features in ASD risk prediction. Random Forest and XGBoost models are utilised to assess feature importance, offering insights into the relative contribution of various clinical, genetic, and environmental variables. Additionally, SHAP values are employed to provide model interpretability and explain the influence of individual features in a transparent and human-understandable manner. Our findings indicate that these AI-driven methods can effectively identify key risk factors for ASD, providing insights that may support earlier detection, targeted prevention, and personalised treatment strategies. This paper underscores the potential of AI to transform autism research through scalable, interpretable, and data-driven methodologies.