EEG–EMG simulation data for caCOH-based hypergraphs

This archive contains synthetic EEG and EMG time-series data generated for the evaluation of graph- and hypergraph-based representations of multivariate neural activity. Dataset overview Each simulation contains: synthetic EEG data from 61 channels; synthetic EMG data from 10 channels. The dataset includes simulations with three true signal-generating dipoles. The three target frequency bands are: 10–12 Hz; 22–24 Hz; 34–36 Hz. The EEG signal-to-noise ratio was varied across eight levels: 0.001; 0.002; 0.005; 0.01; 0.02; 0.05; 0.1; 0.2. For each EEG SNR level, 100 independent repetitions are provided. In total, the archive contains 800 EEG files and 800 EMG files. File structure The archive is organized as follows: dataset_root/ eeg/ eeg_*.npy emg/ emg_*.npy The files in eeg/ and emg/ follow the same naming convention: eeg_SSSS_3_RRR.npyemg_SSSS_3_RRR.npy where: SSSS is the EEG SNR tag multiplied by 1000 and padded to four digits; 3 indicates that the simulation contains three true signal-generating dipoles; RRR is the repetition number, from 001 to 100. For example: eeg_0020_3_015.npyemg_0020_3_015.npy corresponds to EEG SNR = 0.02, three true dipoles, and repetition 15. Data format The EEG and EMG signals are stored as NumPy arrays in .npy format. Each EEG file contains an array with shape (61, 20000), corresponding to 61 EEG channels and 20,000 time points. Each EMG file contains an array with shape (10, 20000), corresponding to 10 EMG channels and 20,000 time points. The sampling frequency is 200 Hz, and each simulation has a duration of 100 seconds. Signal values are stored in volts. Related publication A more detailed description of the data generation procedure, simulation design, and intended analysis is provided in the associated manuscript: Daniil Vlasenko, Irina Saranskaia, Denis Zakharov, Hypergraphs from multivariate connectivity: caCOH-based EEG/MEG representation. arXiv:2606.01357, 2026. https://doi.org/10.48550/arXiv.2606.01357 Citation If you use this dataset in your research, please cite both this archive and the associated manuscript.