Analysis And Detection of Autism Spectrum Disorder Using Multimodel Machine Learning Techniques

Abstract

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition characterized by difficulties in social communication and repetitive behavioural patterns that affect cognitive development. The increasing prevalence of ASD highlights the need for reliable and early screening systems. Traditional diagnostic methods depend on clinical observation and structured assessments, which may delay early identification. This implementation presents a multimodal machine learning framework that integrates prenatal biological indicators and early childhood behavioural features for ASD risk prediction. Random Forest, Artificial Neural Network, and Long Short-Term Memory models were implemented independently and combined using a weighted ensemble strategy. Experimental results demonstrate that prenatal models achieved higher classification performance compared to behavioral models, with Artificial Neural Network producing the highest accuracy. The ensemble model improved prediction stability and reduced false-negative rates, showing that multimodal machine learning integration can effectively support early ASD detection systems.