Multimodal Dissection of the Autism Spectrum

Explore a comprehensive research presentation that demonstrates how data-driven approaches can transform autism diagnosis and treatment by identifying distinct clinical subtypes within the autism spectrum. Learn about a groundbreaking framework that combines massive electronic health record analytics with deep-phenotyped cohort studies to address autism's profound clinical and genetic heterogeneity through precision medicine approaches. Discover how researchers analyzed 63,673 children with autism from a national EHR repository using advanced machine learning techniques including dimensionality reduction (PaCMAP) and density-based clustering (HDBSCAN) to identify 36 distinct clinical clusters with interpretable comorbidity signatures spanning neurologic, developmental, psychiatric, and metabolic profiles. Understand the development of a gradient-boosting prediction model (LightGBM) that can assign new patients to these clusters with high accuracy (AUC 0.9) at their first diagnostic encounter, enabling early and individualized intervention trajectories. Examine mechanistically informative case studies including a dyslipidemia-associated autism subtype identified through multi-dimensional genomic analyses of 3,531 individuals, revealing convergence of autism-segregating deleterious variants within lipid regulation pathways and affecting approximately 5% of children with autism. Investigate findings from the Simons Simplex Collection regarding autistic children who show marked behavioral improvements during febrile episodes, linking fever responsiveness to maternal infection during pregnancy and gastrointestinal dysfunction through immune-gut interactions. Analyze diagnostic timing challenges using data from 23,632 SPARK cohort participants, addressing why 25% of children receive autism diagnoses after age six and identifying two distinct late-diagnosed groups: those characterized by camouflaging behaviors and others affected by clinical overshadowing where severe comorbidities mask underlying autism symptoms.