From Pairwise to High-Order: Hypergraph Methods and Multivariate Connectivity Metrics for EEG/MEG

This paper focuses on hypergraph-based approaches for analyzing functional brain connectivity, where vertices represent EEG/MEG sensors or brain regions, and hyperedges are formed using multidimensional connectivity metrics. We compare classical pairwise connectivity metrics with more advanced multivariate connectivity metrics that offer deeper insights into brain dynamics. Existing hypergraph construction methods and their advantages and limitations are discussed. We propose new algorithms that incorporate both neural EEG/MEG signals and external signals (e.g., electromyography (EMG), electrooculography (EOG), stimulus channels). These methods reveal previously unnoticed potential for revealing more complex patterns and dynamic features of brain activity and promise to enhance EEG/MEG analysis in both research and clinical contexts.