| Issue |
ITM Web Conf.
Volume 72, 2025
III International Workshop on “Hybrid Methods of Modeling and Optimization in Complex Systems” (HMMOCS-III 2024)
|
|
|---|---|---|
| Article Number | 04008 | |
| Number of page(s) | 9 | |
| Section | Data Mining, Machine Learning and Pattern Recognition | |
| DOI | https://doi.org/10.1051/itmconf/20257204008 | |
| Published online | 13 February 2025 | |
EEG-based epileptic seizure detection with a bidirectional long short-term memory deep learning model
1 Siberian Federal University, 79, Svobodny pr., 660041, Krasnoyarsk, Russia
2 Reshetnev Siberian State University of Science and Technology, 31, Krasnoyarskiy Rabochiy pr., 660037, Krasnoyarsk, Russia
* Corresponding author: egorova_ld@rambler.ru
This paper presents a method for detecting epileptic seizures based on electroencephalogram (EEG) analysis using a deep learning model based on Bidirectional Long Short-Term Memory (BiLSTM). The proposed model architecture allows taking into account temporal dependencies and nonlinear dynamics of EEG signals, which makes it effective for recognizing patterns associated with epileptic seizures. The model uses frequency, dynamic, fractal, correlation and statistical characteristics of the EEG signal as informative features. The study includes the stages of data preprocessing, feature extraction and neural network training. To improve the accuracy of the model, data normalization and regularization methods were used. The experimental results obtained on the publicly available TUH EEG dataset demonstrate high performance of the model in detecting epileptic activity: Sensitivity 96.2, Specificity 99.8, F1-score 0.77, AUC 0.98.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.