From Dyadic to Higher-Order Interactions: Enhanced Representation of Homotopic Functional Connectivity Through Control of Intervening Variables.

Abstract

Background: The brain's complex functionality emerges from network interactions that go beyond dyadic connections, with higher-order interactions significantly contributing to this complexity. Homotopic functional connectivity (HoFC) is a key neurophysiological characteristic of the human brain, reflecting synchronized activity between corresponding regions in the brain's hemispheres. Materials and Methods: Using resting-state functional magnetic resonance imaging data from the Human Connectome Project, we evaluate dyadic and higher-order interactions of three functional connectivity (FC) parameterizations-bivariate correlation, partial correlation, and tangent space embedding-in their effectiveness at capturing HoFC through the inter-hemispheric analogy test. Results: Higher-order feature vectors are generated through node2vec, a random walk-based node embedding technique applied to FC networks. Our results show that higher-order feature vectors derived from partial correlation most effectively represent HoFC, while tangent space embedding performs best for dyadic interactions. Discussion: These findings validate HoFC and underscore the importance of the FC construction method in capturing intrinsic characteristics of the human brain.