Cross-scale spatially-aware generative modeling of transcriptomic programs underlying neurodegenerative brain organization.

Neurodegenerative disorders such as Alzheimer's disease exhibit highly organized patterns of regional brain vulnerability, yet the biological mechanisms underlying this spatial selectivity remain incompletely understood. While previous imaging-transcriptomic studies have primarily focused on correlation-based analyses between gene expression and neuroimaging phenotypes, these approaches often lack generative biological representations capable of modeling how transcriptomic organization gives rise to large-scale neurodegenerative structure. In this study, we introduce a cross-scale spatially-aware generative framework for modeling transcriptomic programs underlying cortical neurodegeneration. Regional transcriptomic profiles were derived from the Allen Human Brain Atlas using 910 landmark genes aggregated across 68 cortical regions. Neurodegenerative vulnerability maps were constructed from ADNI FreeSurfer cortical thickness measurements by computing regional cortical thinning differences between cognitively normal controls (NC = 926) and Alzheimer's disease subjects (AD = 426). A variational generative architecture was then used to learn latent biological programs linking regional gene-expression organization to macroscale cortical degeneration. To preserve biologically plausible spatial organization, the model additionally incorporated graph-based spatial smoothness regularization across neighboring cortical regions. The proposed framework achieved strong prediction of regional neurodegenerative vulnerability, yielding an