e4_cogniload

Installation

the model runs inside a conda environment, the setup of that is described README_modell.md

Information about the machine learning model

detailed information is in README_modell.md, email.txt and fit_modell.ipynb as well as comments in the source code in util/modell.py

To train the model:

• put unzipped empatica files into the dataset folder, in a subfolder named after the participant
• edit the score_and_help_indicator.csv file to include the new participant and the session that was recoreded
• run retrain_modell.py (currently, I dont know if there is any cross correlation between participants, but the more, the longer it takes)

To run the actual calculation

configure the participants to look for in the file cogni_streamer.py and run it. It will look for empatica data on the LSL network of those participatns, calculate their load and publish the result again to LSL

other scripts:

e4_emulator.py reads empatica files in the "dataset" folder and outputs directly, as if the people are wearing the wristband just now.

e4_lowpass.py attempts to read the same stream, lowpass filters the heart rate and returns a sample of hr, gsr and acc vector length.