A comprehensive framework for statistical testing of brain dynamics

Area of Science:

• Neuroscience
• Computational Biology
• Statistical Modeling

Background:

• Analyzing temporal dynamics of neural activity is crucial for understanding brain function.
• Quantifying relationships between neural data and behavioral/physiological variables is challenging.
• Advanced computational modeling is often required for these analyses.

Purpose of the Study:

• To provide a protocol for statistical analysis of brain dynamics.
• To test associations between neural activity and non-imaging variables.
• To offer a user-friendly toolbox for researchers.

Main Methods:

• Utilizes a generalized hidden Markov model (HMM), specifically the Gaussian-linear HMM.
• Employs permutation-based methods and structured Monte Carlo resampling for statistical inference.
• Supports multiple experimental modalities (task-based, resting-state) and handles confounding variables.

Main Results:

• The protocol facilitates the quantification and testing of associations between neural dynamics and other variables.
• The open-source Python package offers both a library and a graphical interface.
• Includes tools for intuitive visualization of statistical results and comprehensive tutorials.

Conclusions:

• The developed protocol covers the complete workflow for statistical analysis of functional neural data.
• The toolbox is accessible to researchers with varying programming expertise.
• Enables robust analysis of brain dynamics in neuroscience and mental health research.