Advancing Early Autism Diagnosis Using Multimodal Neuroimaging and Ai-Driven Biomarkers for Neurodevelopmental Trajectory Prediction

Abstract

Early and accurate diagnosis of Autism Spectrum Disorder (ASD) is crucial for timely intervention and improved developmental outcomes. Traditional diagnostic approaches, primarily reliant on behavioral assessments, often lack objectivity and are limited in detecting early neurobiological changes. This review explores the integration of multimodal neuroimaging techniques—including functional MRI (fMRI), diffusion tensor imaging (DTI), and electroencephalography (EEG)—with artificial intelligence (AI)-driven models to enhance early ASD detection. We examine recent advances in identifying neurobiological biomarkers that reflect atypical brain connectivity, structure, and function in infants and young children at risk for ASD. Furthermore, we assess machine learning frameworks capable of learning complex patterns across imaging modalities to predict neurodevelopmental trajectories. Key findings suggest that combining neuroimaging data with deep learning approaches significantly improves diagnostic precision and holds promise for forecasting individual developmental outcomes. Despite these advancements, challenges such as data heterogeneity, interpretability, and ethical considerations remain. This study underscores the transformative potential of AI-integrated neuroimaging in clinical diagnostics and calls for further longitudinal, multimodal research to validate and translate these tools into practice.