Background: Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by impairments in communication, social interaction, and behaviour. Early diagnosis is critical, as timely intervention can significantly improve developmental outcomes. Conventional diagnostic procedures are time-intensive and require specialist expertise, creating a need for scalable, data-driven tools that support early screening in both clinical and non-clinical settings. Objective: This study aims to develop a robust, machine learning-based system capable of accurately predicting ASD risk in children using behavioural, demographic, and clinical indicators. The goal is to provide an accessible tool that helps families and clinicians identify and plan interventions early. Methods: We collected an open-source ASD screening dataset from Kaggle, which contains behavioural indicators, demographic information, and relevant clinical attributes. Multiple machine learning classifiers were trained and evaluated on this dataset to determine the most effective model for early ASD prediction. Standard evaluation metrics were used to compare overall performance. Results: Random Forest achieved the strongest predictive performance among all evaluated models, demonstrating superior accuracy and screening reliability compared to the other classifiers. These results highlight the potential of machine learning approaches for efficient early ASD risk detection. Conclusion: The Random Forest-based ASD prediction model demonstrates strong potential as a dependable early screening tool, offering high predictive accuracy and consistent performance across key evaluation metrics. This data-driven methodology can assist clinicians and caregivers in identifying at-risk children sooner, enabling prompt assessment and enhancing the likelihood of early therapeutic intervention.