Classifying generalized social anxiety disorder (gSAD) using dynamic community analysis of EEG signals.
This repo contains the scripts needed to process raw EEG files (.edf) into a format that be used by static community (Louvain) and dynamic community (CommDy) algorithms. Scikit-learn machine learning algorithms are the final step used for classification of gSAD.
Prerequisites
- Raw EEG data (.edf files)
- A copy of the CommDy and Louvain codebase
- MATLAB
- Python 2.7 (for running CommDy) AND Python 3 (for running sci-kit learn)
This is a rough outline of the steps taken to produce this analysis.
- Received .EDF files from experiment
- On the brain server, do the following steps.
- Run 'WPLI_mod.m' to convert .edf files into .mat files.
- Use the 'mat2pair.sh' shell script to automatically run the 'convert_to_pair.m' script. This converts .mat files into .txt files
- Use python script 'convert_to_pair.py' to convert the saved .txt matrices into a .pair format
- Separated the patient data into their own folders (needs automation)
- Switch to Pachy server, transfer all data. Run the following steps.
- Organize and put all data into labeled folders
- Run the below steps by running bash Auto.sh
- Run Louvain algorithm on each folder.
- Run CommDy on each folder.*
- Analyze with Rstats script.
- Classify the files titled "_ind_stat_c**.txt' using your preferred method (machine learning, etc.)
*See umbertoDifa's brain-project repository for instructions on how to run the CommDy algorithm.
See also the list of contributors who participated in this project.
- Mathew Yang (python, shell, and machine learning scripts) contact for questions: mathew.d.yang@gmail.com
- Luis Love (python and shell scripts)
- Chayant (CommDy author)
- Mengqi Xing (WLPI.m writer, provided EEG data for me)
- Tanya Berger-Wolf (servers and guidance)
This project is licensed under the MIT License - see the LICENSE.md file for details Acknowledgments