Parsing altered brain connectivity in neurodevelopmental disorders by integrating graph-based normative modeling and deep generative networks

Abstract

Divergent brain connectivity is thought to underlie the behavioral and cognitive symptoms observed in many neurodevelopmental disorders. Quantifying divergence from neurotypical connectivity patterns offers a promising pathway to inform diagnosis and therapeutic interventions. While advanced neuroimaging techniques, such as diffusion magnetic resonance imaging, have facilitated the mapping of the brain’s structural connections, the challenge lies in accurately modeling developmental trajectories within these complex networked structures to create robust neurodivergence markers. In this work, we present the Brain Representation via Individualized Deep Generative Embedding framework, which integrates normative modeling with a bio-inspired deep generative model to create a reference trajectory of connectivity transformation as part of neurotypical development. Using data from 1,164 neurotypical participants from the Philadelphia Neurodevelopmental Cohort (PNC) and the Center for Autism Research (CAR), together with 229 autistic participants from CAR, the framework quantifies neurodivergence by comparing individuals against the established neurotypical trajectory. The framework provides a global neurodivergence score based on the difference between connectivity-based brain age and chronological age, along with region-wise neurodivergence maps that highlight localized connectivity differences. Here we show that application of the framework to a large cohort of children with autism spectrum disorder demonstrates that the global neurodivergence score correlates with clinical assessments in autism, and that the regional maps offer insights into individual-level heterogeneity in neurodevelopmental disorders. Together, the neurodivergence score and maps form powerful tools for quantifying developmental divergence in connectivity patterns. This approach advances the development of imaging markers for personalized diagnosis and intervention in various clinical contexts. Brain development depends on how different brain regions become connected over time. In neurodevelopmental conditions such as autism, these connections may develop differently, which may help explain differences in behavior and thinking. This study aimed to measure how much a child’s brain connectivity differs from typical development. We developed a brain imaging framework that learns patterns of typical brain connectivity across childhood. The framework then compares each child with this typical developmental pattern. It produces one overall score and a brain map showing where connectivity differs most. In children with autism, the overall score is linked to clinical assessments, and the maps show individual patterns of difference. This approach may help researchers better understand autism and support future tools for more personalized care. Shen et al. introduce a brain imaging framework that builds a reference trajectory of neurotypical connectivity development. In children with autism, it identifies global neurodivergence associated with clinical assessments and regional maps that reveal individual differences.

The authors' abstract, as published at the source. Communications Medicine, 2026 · DOI ↗