Hypergraph-based brain network switching rate and its application in epilepsy

The community detection-based method for measuring brain network switching rate is inherently limited in characterizing the complex dynamics of neural processes. This constraint fundamentally originates from its exclusive reliance on pairwise connectivity metrics and inability to capture higher-order interactions. To address this limitation, we propose a hypergraph-based framework for quantifying node switching rate, which provides a conceptual shift in perspective for evaluating brain flexibility. This method models brain regions as nodes and depicts multibody interactions via hyperedges, thereby enabling explicit characterization of higher-order functional dynamics. The core idea lies in calculating the sum of absolute differences in hyperedge types between consecutive time windows, normalized to yield the node switching rate. Additionally, we apply the proposed method to 25 seizure events from 11 patients with temporal lobe epilepsy and compare it with the community detection algorithm. Main results include: (i) The hypergraph-based switching rates are significantly elevated during the ictal phase compared to preictal and postictal phases. (ii) In the hypergraph-based method, the seizure onset zone (SOZ) exhibits intensified dynamic fluctuations in the preictal phase, whereas the propagation zone (PZ) demonstrates heightened switching activity during the ictal phase. Notably, all brain zones, including the non-involved zone (NIZ), show increased functional state transitions during seizures. (iii) The hypergraph-based method outperforms the community detection-based measure in capturing temporal complexity and interactions between brain regions. Collectively, the proposed method provides a novel perspective for investigating brain dynamics and demonstrates potential for addressing epilepsy and other neurological disorders.