Abstract
Reliable seizure prediction can improve patient safety by enabling timely protective actions, yet most high-performing approaches depend on multichannel EEG, limiting feasibility in wearable and low-power environments. This study developed an ultralight deep learning model for seizure prediction using only single-channel EEG and evaluated its clinical applicability under predefined, clinically actionable criteria: a seizure prediction horizon (SPH) of 2 minutes and a seizure occurrence period (SOP) of 30 minutes. A 37,985-parameter MobileNet-derived architecture was designed to process STFT spectrograms and validated using patient-specific leave-one-out cross-validation on the SNUH and CHB-MIT datasets. Performance was assessed using segment-based accuracy and false positive rate (FPR), along with event-based sensitivity computed under SPH–SOP constraints. The model demonstrated strong and consistent performance across both datasets. In the SNUH cohort, it achieved 85.97% accuracy, an FPR of 0.130, and 94.93% sensitivity, successfully predicting 95.08% of seizures within the SOP window. In the CHB-MIT dataset, it reached 90.72% accuracy, an FPR of 0.092, and 97.92% sensitivity. These findings provide the first evidence that single-channel EEG can support reliable seizure prediction within clinically meaningful early-warning windows. The proposed lightweight model delivers multichannel-comparable performance while drastically reducing computational demand, demonstrating strong potential for real-time deployment in wearable and patient-centered epilepsy management systems.
Similar content being viewed by others
Data availability
The CHB-MIT dataset is publicly accessible on PhysioNet. The SNUH EEG dataset is classified as protected clinical data and cannot be shared publicly. External transfer is permitted only with prior IRB approval and institutional authorization, and may be provided upon reasonable request to the corresponding author.
References
Fisher, R. S., Acevedo, C., Arzimanoglou, A. et al. ILAE official report: A practical clinical definition of epilepsy. Epilepsia. 2014;55(4):475-482. https://doi.org/10.1111/epi.12550.Lancet Neurol. 2019;18(4):357–375.
Fisher, R. S. et al. ILAE official report: A practical clinical definition of epilepsy. Lancet Neurol. 18(4), 357–375 (2019).
Kwan, P. & Brodie, M. J. Early identification of refractory epilepsy. N Engl. J. Med. 342(5), 314–319 (2000).
Fisher, R. S. et al. Epileptic seizures and epilepsy: Definitions proposed by the international league against epilepsy (ILAE) and the international bureau for epilepsy (IBE). Epilepsia 46(4), 470–472 (2005).
Kerr, M. P. The impact of epilepsy on patients’ lives. Acta Neurol Scand. 126, 1–9 (2012).
Kerr, W. T., McFarlane, K. N. & Pucci, G. F. The present and future of seizure detection, prediction, and forecasting with machine learning, including the future impact on clinical trials. Front Neurol. 15, 1425490 (2024).
He, C. et al. Diversity and suitability of the state-of-the-art wearable and wireless EEG systems review. IEEE J. Biomed Health Inform. 27(8), 3830–3843 (2023).
Yang, X., Zhao, J., Sun, Q., Lu, J. & Ma, X. An effective dual self-attention residual network for seizure prediction. IEEE Trans. Neural. Syst. Rehabil. Eng. 29, 1604–1613 (2021).
Shi, S. & Liu, W. B2-ViT Net: Broad vision transformer network with broad attention for seizure prediction. IEEE Trans. Neural Syst. Rehabil. Eng. 32, 178–188 (2024).
Hussein, R., Lee, S. & Ward, R. Multi-channel vision transformer for epileptic seizure prediction. Biomedicines 10(7), 1551 (2022).
Daoud, H. & Bayoumi, M. A. Efficient epileptic seizure prediction based on deep learning. IEEE Trans. Biomed. Circuits Syst. 13(5), 804–813 (2019).
Frankel, M. A. et al. Wearable reduced-channel EEG system for remote seizure monitoring. Front Neurol. 12, 728484 (2021).
McKenzie, E. D. et al. Validation of a smartphone-based EEG among people with epilepsy: A prospective study. Sci Rep. 7(1), 45567 (2017).
Lu, L, et al. A seizure prediction method for epilepsy utilizing lightweight convolutional neural networks and time-frequency spectrograms of EEG signals. In 2024 IEEE BioCAS; Xi’an, China. 2024:1-5.
Chung, Y. G., Cho, A., Kim, H. & Kim, K. J. Single-channel seizure detection with clinical confirmation of seizure locations using CHB-MIT dataset. Front Neurol. 15, 1389731 (2024).
Ren, Z., Han, X. & Wang, B. The performance evaluation of the state-of-the-art EEG-based seizure prediction models. Front Neurol. 13, 1016224 (2022).
Ohira, J., Yoshimura, H., Morimoto, T., Ariyoshi, K. & Kohara, N. Factors associated with the duration of the postictal state after a generalized convulsion. Seizure 65, 101–105. https://doi.org/10.1016/j.seizure.2019.01.001 (2019).
Yang, H. et al. An end-to-end seizure prediction approach using long short-term memory network. Front Neurosci. 15, 733473 (2023).
Smith, R.G. et al. Pro-Ictal state in human temporal lobe epilepsy. NEJM Evid. 2022;1(3).
Leal, A. et al. Unsupervised EEG preictal interval identification in patients with drug-resistant epilepsy. Sci Rep. 13, 784 (2023).
Lee, D. H. et al. A ResNet-LSTM hybrid model for predicting epileptic seizures using a pretrained model with supervised contrastive learning. Sci Rep. 14(1), 1319 (2024).
Peng, P., Song, Y., Yang, L. & Wei, H. Seizure prediction in EEG signals using STFT and domain adaptation. Front Neurosci. 15, 825434 (2022).
Howard. A.G., Zhu, M., Chen, B. et al. MobileNets: Efficient convolutional neural networks for mobile vision applications. In 2017. Available from: arXiv: 1704.04861
Truong, N. D. et al. Convolutional neural networks for seizure prediction using intracranial and scalp electroencephalogram. Neural Netw. 105, 104–111 (2018).
Meisel, C. & Bailey, K. A. Identifying signal-dependent information about the preictal state: A comparison across ECoG, EEG and EKG using deep learning. eBioMedicine 45, 422–431. https://doi.org/10.1016/j.ebiom.2019.07.001 (2019).
Peng, G., Nourani, M., Harvey, J. & Dave, H. Personalized EEG feature selection for low-complexity seizure monitoring. Int. J. Neur. Syst. 31(08), 2150018. https://doi.org/10.1142/S0129065721500180 (2021).
Wang, Z., Yang, J., Wu, H., Zhu, J. & Sawan, M. Power efficient refined seizure prediction algorithm based on an enhanced benchmarking. Sci. Rep. 11(1), 23498 (2021).
Teixeira, C., Direito, B., Bandarabadi, M., Dourado, A. Output regularization of SVM seizure predictors: Kalman Filter versus the “Firing Power” method. In IEEE Eng. Med. Biol. Soc. Conf. 2012:6530-6533.
Hadady, L., Robinson, T., Bruno, E., Richardson, M.P., Beniczky, S. Users’ perspectives and preferences on using wearables in epilepsy: A critical review. Epilepsia. 2025;66 Suppl 3(Suppl 3):4-13. https://doi.org/10.1111/epi.18280
Qiu, S., Wang, W. & Jiao, H. LightSeizureNet: A lightweight deep learning model for real-time epileptic seizure detection. IEEE J. Biomed. Health Inform. 27(4), 1845–1856 (2023).
Ghaempour, M., Hassanli, K., Abiri, E. An approach to detect and predict epileptic seizures with high accuracy using convolutional neural networks and single-lead-ECG signal. Biomed. Phys. Eng. Express. 2024 Feb 29;10(2). https://doi.org/10.1088/2057-1976/ad29a3.
Zhang, L., Tiwana, B., Qian, Z., Wang, Z., Dick, R.P., Mao, Z.M., Yang, L. Accurate online power estimation and automatic battery behavior based power model generation for smartphones. In IEEE/ACM/IFIP Int Conf Hard Soft Codesign Syst Synthesis (CODES+ISSS). 2010;:105-114.
Yoon, C., Lee, S., Choi, Y., Ha, R. & Cha, H. Accurate power modeling of modern mobile application processors. J. Syst. Archit. 81, 17–31 (2017).
Rattagan, E., Chu, E. T. H., Lin, Y. D. & Lai, Y. C. SEMI: Semi-online power estimates for smartphone hardware components. IEEE Trans. Sustain. Comput. 1(2), 54–62 (2016).
Acknowledgements
This work was supported by the Korea Health Industry Development Institute (KHIDI) grant funded by the Korea government (MOHW) (No. RS-2023-00266765).
Funding
This work was supported by the Ministry of Health and Welfare of Korea (MOHW) through the Korea Health Industry Development Institute (KHIDI) (Grant No. RS-2023-00266765).
Author information
Authors and Affiliations
Contributions
D.J. conceived and implemented the seizure prediction model, performed EEG preprocessing, training, evaluation, and led the manuscript writing. K.-Y.J. contributed to manuscript writing, supervised the overall clinical component, and provided clinical interpretation. Y.-G.J. organized and curated the EEG datasets. T.-J.K. provided clinical insights and assisted in medical interpretation. S.K.L. reviewed the clinical validity and approved the final version of the manuscript. K.-Y.M. contributed to manuscript revision and writing, led the system design and technical direction, and supervised the project from the engineering side. All authors reviewed and approved the final manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Ethical approval
This study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-2411-051-1585). All procedures were conducted in accordance with relevant guidelines and regulations, including the Declaration of Helsinki and the Bioethics and Safety Act. Informed consent was obtained from all participants or their legal guardians.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Jang, D., Jung, KY., Jeon, YG. et al. Single-channel EEG-based seizure prediction using deep learning. Sci Rep (2026). https://doi.org/10.1038/s41598-026-44670-7
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-026-44670-7
