Skip to content

Latest commit

 

History

History
32 lines (24 loc) · 2.07 KB

README.md

File metadata and controls

32 lines (24 loc) · 2.07 KB

Pain Recognition with EEG

In this project, we are analyzing 24 channels dry EEG data coming from the Shrinner Hospital in two patient population: Healthy and participant with MSK.

Code written by Yacine Mahdid in 2020 from home during the Covid-19 pandemic.

Table of Content

Labels

  • rest: is when the participant didn't move while we recorded the data. This was not a very good control since in the hot1/hot2 condition they are moving a cursor
  • nopain: is the same thing if a participant has it we need to change it to rest
  • covas: is when we told the participant to move the small cursor (this is the better control)
  • hot1: the first painful thermal condition
  • cold: the only cold bath condition
  • hot2: the second painful thermal condition after the cold bath

Code Structure

  • .doc: where we keep some important documents that the analysis is based on
  • milestones: where we keep the different iteration of the codebase, in synch with Github milestones
    • .data: contains data we need for the analysis that are static
    • 0_first_cross_validated_ml_model: its the very first iteration of the machine learning pipeline we did to validate that there was something to do with the data
    • 1_neurips_2020_abstract_submission: This was the modification we brought in for the neurips 2020 asbtract submission. We made the analysis a bit more robust.
    • 2_neurips_2020_paper_submission: A week after the abstract submission for neurips we made some more modification to improve the analysis some more.
    • 3_first_draft_paper: Because of the neurips sumission we had a good start for a paper, we gained some ideas of what to change to make a full papper out of our analysis.
  • utils is where utility function (like plotting or reordering) are stored

If you want to reproduce the latest result you need to use the latest milestone

Starting at milestone 3 we decided to use a JSON file for holding the configuration since at that time we were using MATLAB and Python for the analysis and there was state shared between the two.