The role of machine learning in autism spectrum disorder assessment and management

Autism Spectrum Disorder (ASD) presents significant challenges in diagnosis and treatment, driven by its heterogeneous nature and complex aetiology. Recent advances in machine learning (ML) have facilitated exploration of novel approaches to ASD detection, stratification, and intervention opportunities. This narrative review explores the current ML and artificial intelligence (AI) research landscape across several key domains, including early ASD screening, phenotypic stratification, diagnostic biomarkers, neuroimaging, personalised therapies, and the role of automation and robotics in the treatment of this complex condition. Detailed analyses of these approaches emphasise the transformative but not yet realised potential of ML to improve outcomes for individuals with ASD. IMPACT: Highlights emerging trends, including multimodal AI integration, digital phenotyping, and use of AI to achieve biomarker-driven precision medicine. Provides first comprehensive synthesis of AI advancements in screening, diagnosis and treatment of ASD. Identifies current gaps in AI ASD research, such as dataset heterogeneity, validation issues, and clinical trust barriers.