癲癇預測為一發展中的治療方法，透過預測即將到來的癲癇發作，提早預知患者處置。癲癇預測將可幫助因目前醫療技術限制而無法有效治療之癲癇患者。癲癇預測方式主要以EEG訊號，擷取其訊號中特徵並找出癲癇發作時的差異，進而預測癲癇發作。藉由EEG不同頻道於大腦不同區域所量測的訊號，可找出大腦不同區域間的連結與交互活動，稱為功能性連結，本研究以定向轉移函數(Directed transfer function, DTF)計算頻道訊號間的功能性連結，DTF可呈現頻域、雙向的頻道間相關性，藉由觀察癲癇發作前期與癲癇未發作時期的差別進行預測。近年隨著機器學習發展，使用機器學習模型，自動學習特徵與目標間的關係可有效幫助預測，我們比較兩分類器：L1正則化線性迴歸(Least Absolute Shrinkage and Selection Operator, Lasso)與卷積神經網路(Convolutional neural network, CNN)對於DTF用於癲癇預測的成效，CNN在先前研究，相同場景下有良好的預測效果，而Lasso則在公開癲癇預測比賽中有傑出的預測成果，CNN為非線性分類器，而Lasso為線性分類器，兩者各有其使用上的優缺點。我們以三組公開資料集進行計算和比較，得出在Lasso與CNN上預測準確率並無顯著差異(T-Test p=0.3)，但因Lasso為線性迴歸模型具可解釋性、訓練時長為CNN的0.5%，相對使用計算量少，以及參考Occam's razor原則，建議以Lasso作為以DTF為特徵之癲癇預測用機器學習模型。

Seizure prediction to warn patients of upcoming seizures is aimed at improving the patients’ daily lives. Seizures can be predicted by analyzing electroencephalogram (EEG) signals, which demonstrate functional connectivity or interactions between different areas of the brain. This work analyzes and predicts seizures using the directed transfer function that provides features for Lasso and CNN. The prediction performances of the Lasso and CNN models are AUCs of 0.6 and 0.66, respectively, which are significantly better than random predictions. The results show that the proposed algorithm is effective for seizure prediction.

[1] G. Aggarwal and T. K. Gandhi, “Prediction of Epileptic Seizures Based on Mean Phase Coherence,” Neuroscience Nov. 2017.
[2] K. Lehnertz et al., “State-of-the-Art of Seizure Prediction,” J. Clin. Neurophysiol., vol. 24, no. 2, pp. 147–153, Apr. 2007.
[3] H. Chen et al., “An Explainable Statistical Method for Seizure Prediction Using Brain Functional Connectivity from EEG,” Comput. Intell. Neurosci., vol. 2022, pp. 1–8, Dec. 2022.
[4] B. Pohlmann-Eden and M. Newton, “First Seizure: EEG and Neuroimaging Following an Epileptic Seizure: EEG and Neuroimaging after a First Seizure,” Epilepsia, vol. 49, pp. 19–25, Jan. 2008.
[5] B. Litt and K. Lehnertz, “Seizure Prediction and the Preseizure Period,” Curr. Opin. Neurol., vol. 15, no. 2, pp. 173–177, Apr. 2002.
[6] M. Natu, M. Bachute, S. Gite, K. Kotecha, and A. Vidyarthi, “Review on Epileptic Seizure Prediction: Machine Learning and Deep Learning Approaches,” Comput. Math. Methods Med., vol. 2022, pp. 1–17, Jan. 2022.
[7] 黃宇德, “即時性癲癇預測系統實做使用卷積神經網路及其系統晶片設計,” 碩士論文, 電子研究所, 國立交通大學, 台灣, 2019.
[8] L. Chisci et al., “Real-Time Epileptic Seizure Prediction Using AR Models and Support Vector Machines,” IEEE Trans. Biomed. Eng., vol. 57, no. 5, pp. 1124–1132, May 2010.
[9] P.-N. Yu, C. Y. Liu, C. N. Heck, T. W. Berger, and D. Song, “A Sparse Multiscale Nonlinear Autoregressive Model for Seizure Prediction,” J. Neural Eng., vol. 18, no. 2, p. 026012, Apr. 2021.
[10] P. van Mierlo et al., “Functional Brain Connectivity from EEG in Epilepsy: Seizure Prediction and Epileptogenic Focus Localization,” Prog. Neurobiol., vol. 121, pp. 19–35, Oct. 2014.
[11] A. M. Bastos and J.-M. Schoffelen, “A Tutorial Review of Functional Connectivity Analysis Methods and Their Interpretational Pitfalls,” Front. Syst. Neurosci., vol. 9, Jan. 2016.
[12] F. Mormann, T. Kreuz, R. G. Andrzejak, P. David, K. Lehnertz, and C. E. Elger, “Epileptic Seizures are Preceded by a Decrease in Synchronization,” Epilepsy Res., vol. 53, no. 3, pp. 173–185, Mar. 2003.
[13] M. Hejazi and A. Motie Nasrabadi, “Prediction of Epilepsy Seizure from Multi-channel Electroencephalogram by Effective Connectivity Analysis Using Granger Causality and Directed Transfer Function Methods,” Cogn. Neurodyn., vol. 13, no. 5, pp. 461–473, Oct. 2019.
[14] G. Wang et al., “Seizure Prediction Using Directed Transfer Function and Convolution Neural Network on Intracranial EEG,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 28, no. 12, pp. 2711–2720, Dec. 2020.
[15] B. H. Brinkmann et al., “Crowdsourcing Reproducible Seizure Forecasting in Human and Canine Epilepsy,” Brain J. Neurol., vol. 139, no. Pt 6, pp. 1713–1722, Jun. 2016.
[16] C. W. J. Granger, “Investigating Causal Relations by Econometric Models and Cross-spectral Methods,” Econometrica, vol. 37, no. 3, pp. 424–438, 1969.
[17] K. J. Blinowska-Cieślak and J. Zygierewicz, “Practical Biomedical Signal Analysis Using MATLAB,” CRC Press, 2021, https://books.google.com.tw/books?id=qJ56zgEACAAJ.
[18] L. Kuhlmann et al., “Epilepsyecosystem.org: Crowd-sourcing Reproducible Seizure Prediction with Long-term Human Intracranial EEG,” Brain, Aug. 2018.
[19] A. Shoeb, “Application of Machine Learning to Epileptic Seizure Onset Detection and Treatment,” PhD Thesis, Electrical and Medical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA, 2009.
[20] G. James, D. Witten, T. Hastie, and R. Tibshirani, “An Introduction to Statistical Learning: with Applications in R,” Springer, New York, NY, USA, 2023