Modelate Data

Compute frequency and time-frequency domain connectivity measures. The different available connectivity measures in EEGraph with their default parameters.

Connectivity Measures:

• Cross Correlation
  • 'cross_correlation'
    • Threshold = 0.5
    • Does not require frequency bands.
• Pearson Correlation
  • 'pearson_correlation'
    • Threshold = 0.7
    • Does not require frequency bands.
• Squared Coherence
  • 'squared_coherence'
    • Threshold = 0.65
    • Requires frequency bands.
• Imaginary Coherence
  • 'imag_coherence'
    • Threshold = 0.4
    • Requires frequency bands.
• Corrected Cross Correlation
  • 'corr_cross_correlation'
    • Threshold = 0.1
    • Does not require frequency bands.
• Weighted Phase Lag Index (WPLI)
  • 'wpli'
    • Threshold = 0.45
    • Requires frequency bands.
• Phase Locking Value (PLV)
  • 'plv'
    • Threshold = 0.8
    • Requires frequency bands.
• Phase Lag Index (PLI)
  • 'pli'
    • Threshold = 0.1
    • Does not require frequency bands.
  • 'pli_bands'
    • Threshold = 0.1
    • Requires frequency bands.
• Directed Transfer Function (DTF)
  • 'dtf'
    • Threshold = 0.3
    • Requires frequency bands.
• Power Spectrum
  • 'power_spectrum'
    • Threshold = includes the percentage highest connectivity value channels (default 0.25=25%).
    • Requires frequency bands.
• Spectral Entropy
  • 'spectral_entropy'
    • Threshold = includes the percentage highest connectivity value channels (default 0.25=25%).
    • Requires frequency bands.
• Shannon Entropy
  • 'shannon_entropy'
    • Threshold = includes the percentage highest connectivity value channels (default 0.25=25%).
    • Does not require frequency bands.

---

 graphs, connectivity_matrix = G.modelate(window_size, connectivity, bands = [None], threshold = [None])

Parameters:

• window_size: Required | int or list

  Desired window size, can be an integer, e.g.(2) or time intervals, e.g. [0,2].

• connectivity: Required | str

  Connectivity measure to compute. One of the listed in the section above.

• bands: Required | list of str

  It is only required if connectivity measure needs it. Desired frequency bands to be used in the analysis, e.g. ['delta','theta','alpha','beta','gamma'].

• threshold: Optional - Default = Specific to Connectivity Measure | float

  User specific threshold used to determine if an edge exists within two channels from the connectivity value.

Returns:

• graphs: dict of NetworkX graph-like objects

  Dictionary containing all the generated NetworkX graphs.

• connectivity_matrix: numpy.ndarray (G x N x N)

  Connectivity matrices from resulting graphs. G is the number of graphs and N is the number of channels.

---

Example without Frequency Bands:

   G = eegraph.Graph()
   G.load_data(path= "eeg_sample_1.edf", exclude = ['EEG TAntI1-TAntI', 'EEG TAntD1-TAntD', 'EEG EKG1-EKG2'])`
   graphs, connectivity_matrix = G.modelate(window_size = 2, connectivity = 'pearson_correlation')

Example with Frequency Bands:

   G = eegraph.Graph()
   G.load_data(path= "eeg_sample_1.edf", exclude = ['EEG TAntI1-TAntI', 'EEG TAntD1-TAntD', 'EEG EKG1-EKG2'])`
   graphs, connectivity_matrix = G.modelate(window_size = 2, connectivity = 'squared_coherence', bands = ['delta','theta','alpha'])