Skip to content

Automated seizure detection coupled with manual validation

License

Notifications You must be signed in to change notification settings

neurosimata/seizy_ml

Repository files navigation

SeizyML

🐍 SeizyML uses interpretable machine learning models to detect 🕵️ seizures from EEG recordings. After the seizures are detected, they can be manually verified with a user-friendly GUI.

  • Currently supports .h5 files as inputs.

How to install

  1. Download and install miniconda on your platform

  2. Clone or Download SeizyML

  3. Start Anaconda's shell prompt, navigate to /seizy_ml and create conda environment:

     conda env create -f environment.yml
    

Launch CLI

Via Anaconda's shell prompt

    # navigate to *seizy* folder
    cd ./seizy_ml
    
    # enter conda environment
    conda activate seizyml

    # Launch CLI
    python cli.py

Configuration settings and file preparation

For configuration settings and file preparation check the configuration document.


How to Use

  • Before using SeizyML for seizure detection a model should be first trained on ground truth (hand-scored) data.
  • To train a model follow the guidelines in the training tutorial.
  • Check out the tutorial doc for a quick guide to using SeizyML for seizure detection.