STS-HGCN-AL

Original PyTorch implementation of "Spatio-Temporal-Spectral Hierarchical Graph Convolutional Network With Semisupervised Active Learning for Patient-Specific Seizure Prediction" (IEEE Transactions on Cybernetics 2021).

Paper: https://ieeexplore.ieee.org/document/9440862

Requirements

The code was implemented using Python 3.8.3 and the following packages:

torch==1.4.0
numpy==1.18.5
scipy==1.5.0

Datasets

STS-HGCN with an active preictal interval learning scheme is evaluated on one public dataset with 19 patients with intractable seizures:

CHB-MIT dataset

Main Results

Citations

If you find the paper or this repo useful, please cite:

@ARTICLE{9440862,
  author={Li, Yang and Liu, Yu and Guo, Yu-Zhu and Liao, Xiao-Feng and Hu, Bin and Yu, Tao},
  journal={IEEE Transactions on Cybernetics}, 
  title={Spatio-Temporal-Spectral Hierarchical Graph Convolutional Network With Semisupervised Active Learning for Patient-Specific Seizure Prediction}, 
  year={2021},
  volume={},
  number={},
  pages={1-16},
  doi={10.1109/TCYB.2021.3071860}}

Contacts