SeizyML

:snake: SeizyML uses interpretable machine learning models to detect :detective: 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.