Abstract

Autism is behavioral consensus diagnostic label based on early difficulties in social-communication and restricted repetitive behaviors domains. However, the singular nature of the label of ‘autism’ masks a large degree of multi-scale heterogeneity within the spectrum and does not equitably balance differing perspectives within autism research. In this talk, I will discuss a new research-oriented framework that proposes a first-level split of autism into a type I versus type II distinction based on non-core language, intellectual, motor, and adaptive functioning features in early development. These subtypes are identifiable in a highly replicable manner with data-driven machine learning techniques and take very different developmental paths leading to highly differentiated later-life outcomes. Neuroimaging and gene expression data point towards differential biological mechanisms between the subtypes that likely begin in early prenatal development and relate to neurogenesis and cortical patterning. Finally, trajectories through early behavioral intervention also differ between these subtypes. In summary, the proposed research-oriented stratification framework presented in this talk will help balance differing perspectives and facilitate a more precise and individualized understanding of biology, outcomes, and responses to intervention in the autisms.