Code for the paper Staging Epileptogenesis with Deep Neural Networks

Published in Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics

The architecture was inspired by the ResNet developed for Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network

Dependecies

The code was originally developed in tensorflow 1.8X. One need to adapt the code to tensorflow 2.0+

File organization

• exp_params.json: parameters for running the experiment, including file path, batch size, epochs, etc.,
• model_params.json: parameters for building models such as number of layers, kernel sizes, dropout rate, etc.
• main_EPG_classification.py: the main file to run the experiment.
• train.py: detailed training procedure including training, testing steps.
• dataio_EPG.py: helper functions regarding data loading, IO reading and writing, etc.
• plots_EPG.py: plot-related helper functions.

Data organization

How to use

python main_EPG_classification.py