In recent years, the classification of electroencephalogram (EEG) signals has attracted considerable interest, particularly in clinical and cognitive neuroscience applications. However, the problem of class imbalance often undermines the performance of classification algorithms and leads to biased predictions in favor of the majority class. In this paper, I provide a review of data approaches that address the class imbalance problem in EEG classification. I explored methods such as oversampling, undersampling, and the application of advanced algorithms such as Generative Adversarial Networks (GANs), which generally showed better overall performance in dealing with class imbalance. In this paper, I reviewed work published in the last 5 years to evaluate the current state of the art in imbalance. In addition, I provide insight into potential gaps in EEG classification, especially in traumatic brain injury (TBI), and discuss the potential of data augmentation based on deep learning, i.e., GANs, which have shown promise in generating synthetic data to improve minority class representation in other application domains for the EEG-based classification model. By highlighting gaps in existing EEG classification methods and proposing GANs as a solution to class imbalance, I aim to contribute to the development of more robust prediction models that may lead to improved patient outcomes in TBI cases.
Citation: Nor Safira Elaina Mohd Noor. Addressing class imbalance in traumatic brain injury prognostication: A survey of resampling approaches[J]. AIMS Neuroscience, 2026, 13(1): 171-207. doi: 10.3934/Neuroscience.2026008
In recent years, the classification of electroencephalogram (EEG) signals has attracted considerable interest, particularly in clinical and cognitive neuroscience applications. However, the problem of class imbalance often undermines the performance of classification algorithms and leads to biased predictions in favor of the majority class. In this paper, I provide a review of data approaches that address the class imbalance problem in EEG classification. I explored methods such as oversampling, undersampling, and the application of advanced algorithms such as Generative Adversarial Networks (GANs), which generally showed better overall performance in dealing with class imbalance. In this paper, I reviewed work published in the last 5 years to evaluate the current state of the art in imbalance. In addition, I provide insight into potential gaps in EEG classification, especially in traumatic brain injury (TBI), and discuss the potential of data augmentation based on deep learning, i.e., GANs, which have shown promise in generating synthetic data to improve minority class representation in other application domains for the EEG-based classification model. By highlighting gaps in existing EEG classification methods and proposing GANs as a solution to class imbalance, I aim to contribute to the development of more robust prediction models that may lead to improved patient outcomes in TBI cases.
| [1] |
Hale AT, Stonko DP, Wang L, et al. (2018) Machine learning analyses can differentiate meningioma grade by features on magnetic resonance imaging. Neurosurg Focus 45: E4. https://doi.org/10.3171/2018.8.FOCUS18191 
|
| [2] |
Hale AT, Stonko DP, Lim J, et al. (2018) Using an artificial neural network to predict traumatic brain injury. J Neurosurg Pediatr 23: 219-226. https://doi.org/10.3171/2018.8.PEDS18370
|
| [3] |
Vespa PM, Boscardin WJ, Hovda DA, et al. (2002) Early and persistent impaired percent alpha variability on continuous electroencephalography monitoring as predictive of poor outcome after traumatic brain injury. J Neurosurg 97: 84-92. https://doi.org/10.3171/jns.2002.97.1.0084
|
| [4] |
Hannawi Y, Stevens RD (2016) Mapping the connectome following traumatic brain injury. Curr Neurol Neurosci Rep 16: 44. https://doi.org/10.1007/s11910-016-0642-9
|
| [5] |
Ianof JN, Anghinah R (2017) Traumatic brain injury: An EEG point of view. Dement Neuropsychol 11: 3-5. https://doi.org/10.1590/1980-57642016dn11-010002
|
| [6] |
Mohd Noor NSE, Ibrahim H (2020) Machine learning algorithms and quantitative electroencephalography predictors for outcome prediction in traumatic brain injury: A systematic review. IEEE Access 8: 102075-102092. https://doi.org/10.1109/ACCESS.2020.2998934
|
| [7] |
Bai W (2024) The predicative value of early quantitative electroencephalograph in epilepsy after severe traumatic brain injury in children. Front Pediatr 12: 1370692. https://doi.org/10.3389/fped.2024.1370692
|
| [8] |
Wang J, Huang L, Ma X, et al. (2022) Role of quantitative EEG and EEG reactivity in traumatic brain injury. Clin EEG Neurosci 53: 452-459. https://doi.org/10.1177/1550059420984934
|
| [9] |
Hyder AA, Wunderlich CA, Puvanachandra P, et al. (2007) The impact of traumatic brain injuries: a global perspective. NeuroRehabilitation 22: 341-353. https://doi.org/10.3233/NRE-2007-22502
|
| [10] |
Lee H, Mizrahi MA, Hartings JA, et al. (2019) Continuous electroencephalography after moderate to severe traumatic brain injury. Crit Care Med 47: 574-582. https://doi.org/10.1097/CCM.0000000000003639
|
| [11] |
Tolonen A, Särkelä MOK, Takala RSK, et al. (2018) Quantitative EEG parameters for prediction of outcome in severe traumatic brain injury: Development study. Clin EEG Neurosci 49: 248-257. https://doi.org/10.1177/1550059417742232
|
| [12] |
van den Brink RL, Nieuwenhuis S, van Boxtel GJM, et al. (2018) Task-free spectral EEG dynamics track and predict patient recovery from severe acquired brain injury. Neuroimage Clin 17: 43-52. https://doi.org/10.1016/j.nicl.2017.10.003
|
| [13] |
Pease M, Elmer J, Shahabadi AZ, et al. (2023) Predicting posttraumatic epilepsy using admission electroencephalography after severe traumatic brain injury. Epilepsia 64: 18421852. https://doi.org/10.1111/epi.17622
|
| [14] |
Zelig D, Goldberg I, Shor O, et al. (2022) Paroxysmal slow wave events predict epilepsy following a first seizure. Epilepsia 63: 190-198. https://doi.org/10.1111/epi.17110
|
| [15] |
Noor NSEM, Ibrahim H, Lah MHC, et al. (2021) Improving outcome prediction for traumatic brain injury from imbalanced datasets using RUSBoosted trees on electroencephalography spectral power. IEEE Access 9: 121608-121631. https://doi.org/10.1109/ACCESS.2021.3109780
|
| [16] | Ellethy H, Chandra SS, Vegh V (2024) Machine learning applications in traumatic brain injury diagnosis and prognosis: A spotlight on mild TBI and CT imaging. arXiv preprint . |
| [17] |
Habashi AG, Azab AM, Eldawlatly S, et al. (2023) Generative adversarial networks in EEG analysis: an overview. J Neuroeng Rehabil 20: 40. https://doi.org/10.1186/s12984-023-01169-w
|
| [18] |
Lashgari E, Liang D, Maoz U (2020) Data augmentation for deep-learning-based electroencephalography. J Neurosci Methods 346: 108885. https://doi.org/10.1016/j.jneumeth.2020.108885
|
| [19] |
Buda M, Maki A, Mazurowski MA (2018) A systematic study of the class imbalance problem in convolutional neural networks. Neural Netw 106: 249-259. https://doi.org/10.1016/j.neunet.2018.07.011
|
| [20] |
Shorten C, Khoshgoftaar TM (2019) A survey on image data augmentation for deep learning. J Big Data 6: 60. https://doi.org/10.1186/s40537-019-0197-0
|
| [21] | Hartmann KG, Schirrmeister RT, Ball T (2018) EEG-GAN: Generative adversarial networks for electroencephalograhic (EEG) brain signals. arXiv preprint . |
| [22] |
Duffy FH, Hughes JR, Miranda F, et al. (1994) Status of quantitative EEG (qEEG) in clinical practice. Clin Electroencephalogr 25: VI-XXII. https://doi.org/10.1177/155005949402500403
|
| [23] |
Varotto G, Susi G, Tassi L, et al. (2021) Comparison of resampling techniques for imbalanced datasets in machine learning: application to epileptogenic zone localization from interictal intracranial eeg recordings in patients with focal epilepsy. Front Neuroinform 15: 715421. https://doi.org/10.3389/fninf.2021.715421
|
| [24] |
Fan J, Sun C, Chen C, et al. (2020) EEG data augmentation: Towards class imbalance problem in sleep staging tasks. J Neural Eng 17: 056017. https://doi.org/10.1088/1741-2552/abb5be
|
| [25] |
Wongvorachan T, He S, Bulut O (2023) A comparison of undersampling, oversampling, and SMOTE methods for dealing with imbalanced classification in educational data mining. Information 14: 54. https://doi.org/10.3390/info14010054
|
| [26] | Brandt J, Lanzén E A comparative review of SMOTE and ADASYN in imbalanced data classification (2021). |
| [27] | Chawla NV, Lazarevic A, Hall LO, et al. SMOTEBoost: Improving prediction of the minority class in boosting (2003). https://doi.org/10.1007/978-3-540-39804-2_12 |
| [28] |
Liang XW, Jiang AP, Li T, et al. (2020) LR-SMOTE — An improved unbalanced data set oversampling based on K-means and SVM. Knowl Based Syst 196: 105845. https://doi.org/10.1016/j.knosys.2020.105845
|
| [29] |
Xie W, Liang G, Dong Z, et al. (2019) An improved oversampling algorithm based on the samples' selection strategy for classifying imbalanced data. Math Probl Eng 2019: 3526539. https://doi.org/10.1155/2019/3526539
|
| [30] |
Mujahid M, Kına E, Rustam F, et al. (2024) Data oversampling and imbalanced datasets: An investigation of performance for machine learning and feature engineering. J Big Data 11: 87. https://doi.org/10.1186/s40537-024-00943-4
|
| [31] | Dudjak M, Martinović G (2020) In-depth performance analysis of SMOTE-based oversampling algorithms in binary classification. Int J Electr Comput Eng Syst 11: 13-23. https://doi.org/10.32985/ijeces.11.1.2 |
| [32] |
Zhou J, Zhang X, Jiang Z (2021) Recognition of imbalanced epileptic EEG signals by a graph-based extreme learning machine. Wirel Commun Mob Comput 2021: 5871684. https://doi.org/10.1155/2021/5871684
|
| [33] |
Lee T, Kim M, Kim SP (2020) Improvement of P300-based brain–computer interfaces for home appliances control by data balancing techniques. Sensors 20: 5576. https://doi.org/10.3390/s20195576
|
| [34] |
Fan Z, Sohail S, Sabrina F, et al. (2024) Sampling-based machine learning models for intrusion detection in imbalanced dataset. Electronics 13: 1878. https://doi.org/10.3390/electronics13101878
|
| [35] |
Seiffert C, Khoshgoftaar TM, Van Hulse J, et al. (2010) RUSBoost: A hybrid approach to alleviating class imbalance. IEEE Trans Syst Man Cybern Part A Syst Hum 40: 185-197. https://doi.org/10.1109/TSMCA.2009.2029559
|
| [36] | Seiffert C, Khoshgoftaar TM, Van Hulse J, et al. (2008) RUSBoost: Improving classification performance when training data is skewed. 2008 19th international conference on pattern recognition. IEEE 2008: 1-4. https://doi.org/10.1109/ICPR.2008.4761297 |
| [37] |
Kinoshita T, Fujiwara K, Kano M, et al. (2020) Sleep spindle detection using RUSBoost and synchrosqueezed wavelet transform. IEEE Trans Neural Syst Rehabil Eng 28: 390-398. https://doi.org/10.1109/TNSRE.2020.2964597
|
| [38] |
Mohd Noor NSE, Ibrahim H, Lah MHC, et al. (2022) Prediction of recovery from traumatic brain injury with eeg power spectrum in combination of independent component analysis and RUSBoost model. Biomedinformatics 2: 106-123. https://doi.org/10.3390/biomedinformatics2010007
|
| [39] |
Sekkal RN, Bereksi-Reguig F, Ruiz-Fernandez D, et al. (2022) Automatic sleep stage classification: From classical machine learning methods to deep learning. Biomed Signal Process Control 77: 103751. https://doi.org/10.1016/j.bspc.2022.103751
|
| [40] |
Khairy M, Mahmoud TM, El-Hafeez TA (2023) The effect of rebalancing techniques on the classification performance in cyberbullying datasets. Neural Comput Appl 36: 1049-1065. https://doi.org/10.1007/s00521-023-09084-w
|
| [41] | Sowah RA, Agebure MA, Mills GA, et al. (2016) New cluster undersampling technique for class imbalance learning. Int J Mach Learn Comput 6: 205214. https://doi.org/10.18178/ijmlc.2016.6.3.599 |
| [42] | Lv Z, Liu Q (2023) ISODF-ENN: Imbalanced data mixed sampling method based on improved diffusion model and ENN. J Intell Fuzzy Syst 46: 1-15. https://doi.org/10.3233/JIFS-233886 |
| [43] |
Wang AX, Le VT, Trung HN, et al. (2025) Addressing imbalance in health data: Synthetic minority oversampling using deep learning. Comput Biol Med 188: 109830. https://doi.org/10.1016/j.compbiomed.2025.109830
|
| [44] |
Guo H, Diao X, Liu H (2019) Improving undersampling-based ensemble with rotation forest for imbalanced problem. Turk J Electr Eng Comput Sci 27: 1371-1386. https://doi.org/10.3906/elk-1805-159
|
| [45] |
Quan D, Feng W, Dauphin G, et al. (2022) A novel double ensemble algorithm for the classification of multi-class imbalanced hyperspectral data. Remote Sens 14: 3765. https://doi.org/10.3390/rs14153765
|
| [46] | Liu TY (2009) Easyensemble and feature selection for imbalance data sets, 2009 international joint conference on bioinformatics, systems biology and intelligent computing. IEEE 2009: 517-520. https://doi.org/10.1109/IJCBS.2009.22 |
| [47] |
Jiang L, Yuan P, Liao J, et al. (2022) Undersampling of approaching the classification boundary for imbalance problem. Concurr Comput Pract Exp 35: 1-1. https://doi.org/10.1002/cpe.7586
|
| [48] |
Haixiang G, Yijing L, Shang J, et al. (2017) Learning from class-imbalanced data: Review of methods and applications. Expert Syst Appl 73: 220-239. https://doi.org/10.1016/j.eswa.2016.12.035
|
| [49] |
McNerney MW, Hobday T, Cole B, et al. (2019) Objective classification of mTBI using machine learning on a combination of frontopolar electroencephalography measurements and self-reported symptoms. Sports Med Open 5: 14. https://doi.org/10.1186/s40798-019-0187-y
|
| [50] |
George O, Smith R, Madiraju P, et al. (2022) Data augmentation strategies for EEG-based motor imagery decoding. Heliyon 8: e10240. https://doi.org/10.1016/j.heliyon.2022.e10240
|
| [51] |
Pei Y, Luo Z, Yan Y, et al. (2021) Data augmentation: Using channel-level recombination to improve classification performance for motor imagery EEG. Front Hum Neurosci 15: 645952. https://doi.org/10.3389/fnhum.2021.645952
|
| [52] |
Kingma DP, Welling M (2019) An introduction to variational autoencoders. Found Trends Mach Learn 12: 307-392. https://doi.org/10.1561/2200000056
|
| [53] |
Zhao T, Cui Y, Ji T, et al. (2024) VAEEG: Variational auto-encoder for extracting EEG representation. NeuroImage 304: 120946. https://doi.org/10.1016/j.neuroimage.2024.120946
|
| [54] |
Sadegh-Zadeh SA, Fakhri E, Bahrami M, et al. (2023) An approach toward artificial intelligence Alzheimer's disease diagnosis using brain signals. Diagnostics 13: 477. https://doi.org/10.3390/diagnostics13030477
|
| [55] |
Xin Y, Walia E, Babyn P (2019) Generative adversarial network in medical imaging: a review. Med Image Anal 58: 101552. https://doi.org/10.1016/j.media.2019.101552
|
| [56] | Rahmadani S, Subekti A, Haris M (2024) Improving classification performance on imbalanced medical data using generative adversarial network. J Ilmu Komput Inf 17: 9-17. https://doi.org/10.21609/jiki.v17i1.1177 |
| [57] |
Borji A (2022) Pros and cons of GAN evaluation measures: New developments. Comput Vis Image Underst 215: 103329. https://doi.org/10.1016/j.cviu.2021.103329
|
| [58] |
Du X, Wang X, Zhu L, et al. (2024) Electroencephalographic signal data augmentation based on improved generative adversarial network. Brain Sci 14: 367. https://doi.org/10.3390/brainsci14040367
|
| [59] |
Luo Tj, Fan Yc, Chen L, et al. (2020) EEG signal reconstruction using a generative adversarial network with wasserstein distance and temporal-spatial-frequency loss. Front Neuroinform 14: 15. https://doi.org/10.3389/fninf.2020.00015
|
| [60] |
Bao G, Yan B, Tong L, et al. (2021) Data augmentation for EEG-based emotion recognition using generative adversarial networks. Front Comput Neurosci 15: 723843. https://doi.org/10.3389/fncom.2021.723843
|
| [61] |
Hu M, Chen J, Jiang S, et al. (2022) E2SGAN: EEG-to-SEEG translation with generative adversarial networks. Front Neurosci 16: 971829. https://doi.org/10.3389/fnins.2022.971829
|
| [62] |
Fahimi F, Došen S, Ang KK, et al. (2021) Generative adversarial networks-based data augmentation for brain–computer interface. IEEE Trans Neural Netw Learn Syst 32: 4039-4051. https://doi.org/10.1109/TNNLS.2020.3016666
|
| [63] |
Lotte F (2015) Signal processing approaches to minimize or suppress calibration time in oscillatory activity-based brain–computer interfaces. Proc IEEE 103: 871890. https://doi.org/10.1109/JPROC.2015.2404941
|
| [64] |
Truong ND, Kuhlmann L, Bonyadi MR, et al. (2019) Epileptic seizure forecasting with generative adversarial networks. IEEE Access 7: 143999-144009. https://doi.org/10.1109/ACCESS.2019.2944691
|
| [65] |
Tibermacine IE, Russo S, Citeroni F, et al. (2025) Adversarial denoising of EEG signals: a comparative analysis of standard GAN and WGAN-GP approaches. Front Hum Neurosci 19: 1583342. https://doi.org/10.3389/fnhum.2025.1583342
|
| [66] | Zhang X, Chen X, Dong M, et al. (2019) Multi-task generative adversarial learning on geometrical shape reconstruction from EEG brain signals. arXiv preprint . |
| [67] | Aznan NKN, Atapour-Abarghouei A, Bonner S, et al. (2019) Simulating brain signals: creating synthetic EEG data via neural-based generative models for improved SSVEP classification. IEEE 2019: 1-8. https://doi.org/10.1109/IJCNN.2019.8852227 |
| [68] |
Brophy E, Redmond P, Fleury A, et al. (2022) Denoising EEG signals for real-world BCI applications using GANs. Front Neuroergon 2: 805573. https://doi.org/10.3389/fnrgo.2021.805573
|
| [69] |
Zeng P, Fan L, Luo YX, et al. (2024) Task-oriented EEG denoising generative adversarial network for enhancing SSVEP-BCI performance. J Neural Eng 21: 066003. https://doi.org/10.1088/1741-2552/ad8963
|
| [70] |
Yasaka K, Akai H, Abe O, et al. (2018) Deep learning with convolutional neural network for differentiation of liver masses at dynamic contrast-enhanced CT: A preliminary study. Radiology 286: 887-896. https://doi.org/10.1148/radiol.2017170706
|
| [71] |
Xu Y, Wu T, Charlton JR, et al. (2022) GAN training acceleration using fréchet descriptor based coreset. Appl Sci 12: 7599. https://doi.org/10.3390/app12157599
|
| [72] |
Frid-Adar M, Diamant I, Klang E, et al. (2018) GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classification. Neurocomputing 321: 321-331. https://doi.org/10.1016/j.neucom.2018.09.013
|
| [73] |
Hu D, Cao J, Lai X, et al. (2021) Epileptic signal classification based on synthetic minority oversampling and blending algorithm. IEEE Trans Cogn Dev Syst 13: 368-382. https://doi.org/10.1109/TCDS.2020.3009020
|
| [74] |
Alhudhaif A (2021) A novel multi-class imbalanced EEG signals classification based on the adaptive synthetic sampling (ADASYN) approach. PeerJ Comput Sci 7: e523. https://doi.org/10.7717/peerj-cs.523
|
| [75] | Masum M, Shahriar H, Haddad H (2020) Analysis of sampling techniques towards epileptic seizure detection from imbalanced dataset. IEEE 2020: 684-692. https://doi.org/10.1109/COMPSAC48688.2020.0-180 |
| [76] |
Abou-Abbas L, Henni K, Jemal I, et al. (2024) Generative AI with WGAN-GP for boosting seizure detection accuracy. Front Artif Intell 7: 1437315. https://doi.org/10.3389/frai.2024.1437315
|
| [77] |
Ganti B, Chaitanya G, Balamurugan R, et al. (2022) Time-series generative adversarial network approach of deep learning improves seizure detection from the human thalamic SEEG. Front Neurol 13: 755094. https://doi.org/10.3389/fneur.2022.755094
|
| [78] | Kan CNE, Povinelli RJ, Ye DH (2021) Enhancing multi-channel EEG classification with gramian temporal generative adversarial network. IEEE 2021: 1260-1264. https://doi.org/10.1109/ICASSP39728.2021.9414078 |
| [79] |
Shah V, von Weltin E, Lopez S, et al. (2018) The temple university hospital seizure detection corpus. Front Neuroinform 12: 83. https://doi.org/10.3389/fninf.2018.00083
|
| [80] |
Wardoyo R, Wirawan IMA, Pradipta IGA (2022) Oversampling approach using Radius-SMOTE for imbalance electroencephalography datasets. Emerg Sci J 6: 382398.
|
| [81] | Zhang Z, Zhong S, Liu Y (2024) Beyond mimicking under-represented emotions: Deep data augmentation with emotional subspace constraints for EEG-based emotion recognition. Proc AAAI Conf Artif Intell 38: 10252-10260. https://doi.org/10.1609/aaai.v38i9.28891 |
| [82] | Krishnamoorthy V Detection and prediction of microsleeps from EEG via deep learning: University of Otago (2021). |
| [83] |
Alimisis P, Mademlis I, Radoglou-Grammatikis P, et al. (2025) Advances in diffusion models for image data augmentation: A review of methods, models, evaluation metrics and future research directions. Artif Intell Rev 58: 112. https://doi.org/10.1007/s10462-025-11116-x
|
| [84] |
Torma S, Szegletes L (2025) Generative modeling and augmentation of EEG signals using improved diffusion probabilistic models. J Neural Eng 22: 016001. https://doi.org/10.1088/1741-2552/ada0e4
|
| [85] |
You Z, Guo Y, Zhang X, et al. (2025) Virtual electroencephalogram acquisition: A review on electroencephalogram generative methods. Sensors 25: 3178. https://doi.org/10.3390/s25103178
|
| [86] |
Zhang D, Ma M, Xia L (2022) A comprehensive review on GANs for time-series signals. Neural Comput Appl 34: 3551-3571. https://doi.org/10.1007/s00521-022-06888-0
|
| [87] |
Haveman ME, Van Putten MJAM, Hom HW, et al. (2019) Predicting outcome in patients with moderate to severe traumatic brain injury using electroencephalography. Crit Care 23: 401. https://doi.org/10.1186/s13054-019-2656-6
|
| [88] |
Noor NSEM, Ibrahim H, Qin LC, et al. (2025) EEG-based prognostic prediction in moderate traumatic brain injury: A hybrid BiLSTM-AdaBoost approach. IEEE Access 13: 172157-172182. https://doi.org/10.1109/ACCESS.2025.3608067
|
| [89] |
Yang B, Sun X, Shi Q, et al. (2022) Prediction of early prognosis after traumatic brain injury by multifactor model. CNS Neurosci Ther 28: 2044-2052. https://doi.org/10.1111/cns.13935
|
| [90] |
Jiang L, Yin X, Yin C, et al. (2011) Different quantitative EEG alterations induced by TBI among patients with different APOE genotypes. Neurosci Lett 505: 160-164. https://doi.org/10.1016/j.neulet.2011.10.011
|
Figures(3) / Tables(1)
Nor Safira Elaina Mohd Noor. Addressing class imbalance in traumatic brain injury prognostication: A survey of resampling approaches[J]. AIMS Neuroscience, 2026, 13(1): 171-207. doi: 10.3934/Neuroscience.2026008
DownLoad: