Abstract
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterised by challenges in social interaction, communication, and restrictive or repetitive behaviours, prompting global research efforts to better understand its prevalence and underlying mechanisms. Among the key contributors to ASD etiologies are common environmental factors, including prenatal, perinatal, and postnatal influences, which play a significant role in the development of the disorder. This study explores key risk factors associated with ASD through the application of feature selection and ensemble learning techniques, with a particular focus on evaluating the impact of environmental variables. We employed Recursive Feature Elimination to identify and rank the most influential factors contributing to ASD based on their effect on model performance. To improve the accuracy and robustness of predictions, we integrated multiple classifiers within an ensemble learning framework. Our approach enhances the reliability of the findings and offers a scalable, data-driven methodology for future research on the multifactorial nature of ASD risk.