p-Value combiners for graphical modelling of EEG data in the frequency domain

Area of Science:

• Neuroscience
• Biostatistics
• Signal Processing

Background:

• Estimating brain connectivity and statistical significance is crucial for graphical modeling.
• Current methods aggregate results across frequencies and patients to build network models.

Purpose of the Study:

• To introduce a novel p-value combiner method for electroencephalography (EEG) data analysis.
• To improve the accuracy and reliability of brain network modeling.

Main Methods:

• A two-step protocol using p-value combiners: frequency-wide and group-wide tests.
• Application to EEG data from distinct mental health patient groups.
• Comparison with the conventional Holm's Stepdown procedure.

Main Results:

• The proposed method effectively controls the false detection rate in EEG data analysis.
• Graphical models revealed structural connectivity differences between patient groups.
• Reliable network construction with effective control of false connections.

Conclusions:

• The new p-value combiner methodology significantly improves upon the Holm's Stepdown procedure.
• Offers enhanced control over error and false negative rates in network models.
• Demonstrates superior applicability for brain connectivity analysis.