Data-driven lifespan transitions: cortical morphometry and intrinsic differences across network scales

Abstract

Age-related changes in cortical morphology are known to be nonlinear and feature-specific. However, age is often modelled using arbitrary bins or treated as a continuous variable. At the same time, structural covariance networks (SCN) have demonstrated that different morphometric features exhibit intrinsically distinct patterns of network organisation. Here, we integrate these approaches by introducing a data-driven framework to identify lifespan transitions in cortical morphometry and relate them to differences in SCN organisation. Using bootstrap-stabilised decision tree regression, we identify robust age partitions for multiple cortical morphometric features (surface area, its thickness and folding), revealing distinct feature-specific ageing regimes across lifespan (18 - 94 years of age). Our results show that features exhibiting similar lifespan transition profiles also demonstrate similar community-level SCN organisation, while features with divergent age trajectories show different network organisation. These findings demonstrate that lifespan transitions in cortical morphology are intrinsically linked to feature-specific network architecture, supporting the view that morphometric features capture distinct biological processes. Our results highlight the importance of data-driven lifespan modelling and reinforce the need to treat morphometric features as non-interchangeable when constructing network-based models of brain structure.