Higher-order functional connectivity and topological analysis of multichannel fNIRS data

Functional connectivity analysis is widely used to characterize large-scale brain networks in fNIRS, but conventional approaches rely on pairwise interactions and may overlook multiregional coordination. Here, we propose a higher-order functional connectivity and topological analysis framework for multichannel fNIRS data, based on triangle-level co-fluctuations and weighted simplicial-complex representations. Using resting-state fNIRS data, we evaluated the reliability of higher-order connectivity metrics. Triangle-based higher-order connectivity showed consistent spatial-pattern reliability (r = 0.572 intra-session; r = 0.280 inter-session) and moderate feature-wise reliability (single-measure ICC ≈ 0.31-0.32), comparable to pairwise connectivity and higher than edge-centric measures, while topological scaffold features exhibited lower reliability. As an exploratory application, higher-order connectivity demonstrated robust individual identifiability (I _diff = 0.053), exceeding pairwise (0.044) and edge-centric connectivity (0.041), with rank-based identification performance above chance (AUC = 0.58). Scaffold features showed lower I _diff but remained above-chance rank-based identification performance. Overall, these findings demonstrate that higher-order and topological features can be reliably derived from fNIRS data and provide complementary information beyond pairwise connectivity. The proposed framework establishes a methodological foundation for investigating higher-order functional networks in fNIRS.