Epilepsy

Overview

Multivariate time series classification dataset from the UEA archive (Bagnall et al., 2018), based on the data of Villar et al. (2016). Six healthy participants wore a tri-axial accelerometer on the dominant wrist, sampled at 16 Hz, while performing four activities: a mimicked epileptic seizure (trained and controlled following a protocol defined by a medical expert), walking, running and sawing.

The archive removed flat prefixes and suffixes, truncated every recording to the length of the shortest one, 206 steps (about 13 seconds), and dropped one recording that was not collected correctly. The loader passes the archive’s data through unchanged.

Train: 137 recordings from three participants.
Test: 138 recordings from the other three participants.

Class names are the archive’s labels in lower case: epilepsy, running, sawing, walking.

Data Structure

Key	Type	Description
`series`	list of list of float32	Accelerometer recording, shape `(206, 3)`: time steps by channels
`label`	int	Activity class (0-3)

Usage Example

import numpy as np
from stable_datasets.timeseries import Epilepsy

train = Epilepsy(split="train")
sample = train[0]
x = np.asarray(sample["series"])  # (206, 3)
print(x.shape, train.features["label"].names[sample["label"]])

References

Homepage: https://timeseriesclassification.com/description.php?Dataset=Epilepsy
Archive record: https://doi.org/10.5281/zenodo.11206204
Original data: http://di002.edv.uniovi.es/~villar/Jose_R._Villar/Public_Data_sets/Entries/2016/5/3_Seizure_simulation_raw_data.html
License: CC BY 4.0

Citation

@article{villar2016generalized,
  title={Generalized Models for the Classification of Abnormal Movements in Daily Life and its Applicability to Epilepsy Convulsion Recognition},
  author={Villar, José R. and Vergara, Paula and Menéndez, Manuel and de la Cal, Enrique and González, Víctor M. and Sedano, Javier},
  journal={International Journal of Neural Systems},
  volume={26},
  number={6},
  pages={1650037},
  year={2016}
}

@article{bagnall2018uea,
  title={The UEA multivariate time series classification archive, 2018},
  author={Bagnall, Anthony and Dau, Hoang Anh and Lines, Jason and Flynn, Michael and Large, James and Bostrom, Aaron and Southam, Paul and Keogh, Eamonn},
  journal={arXiv preprint arXiv:1811.00075},
  year={2018}
}