Beta-Band Electroencephalography Classification for Autism Spectrum Disorder Using Wavelet Features and Least-Squares Support Vector Machine
Downloads
Autism spectrum disorder requires accessible and objective neurophysiological biomarkers to complement behavioral assessment, particularly for early screening in resource-limited settings. This study explores a computationally efficient framework for distinguishing children with autism spectrum disorder from neurotypical controls using beta-band electroencephalography activity (12–30 Hz), which has been associated with atypical sensorimotor and cognitive processing in autism. Beta-band oscillations are theoretically relevant for their roles in attention, cognitive control, and inhibitory processes, domains frequently disrupted in autism spectrum disorder. Data were obtained from the public King Abdulaziz University dataset comprising 16 male participants (8 with autism, 8 controls; aged 6–14 years). Following independent component analysis-based artifact removal and bandpass filtering, recordings were segmented into 2-s epochs with 50% overlap. Discrete wavelet transform (Daubechies-4, four levels) was applied to extract statistical features (mean, standard deviation, skewness, kurtosis) from wavelet coefficients across 16 EEG channels, yielding a 320-dimensional feature vector per epoch. Classification was performed using least-squares support vector machines with a polynomial kernel (degree d=3), with hyperparameters optimized via 5-fold cross-validation on the training set, and evaluated via a stratified 70/30 train–test split at the segment level. The polynomial-kernel model achieved 98.49% segment-level accuracy, outperforming the linear kernel (95.07%) and a relative beta-power baseline. However, these results should be interpreted with caution due to the small sample size (n=16), a male-only cohort, and segment-level evaluation, which may inflate performance through intra-subject data leakage. The lightweight computational design supports potential implementation on portable devices. This proof-of-concept demonstrates the feasibility of wavelet-based beta-band analysis for autism classification, but rigorous validation using larger, balanced cohorts with subject-wise cross-validation is essential before clinical translation can be considered.
[1] C. Lord, T. S. Brugha, T. Charman, J. Cusack, G. Dumas, T. Frazier, E. Jones, R. M. Jones, A. Pickles, L. W. Robinson, et al., “Autism spectrum disorder,” The Lancet, vol. 392, no. 10146, pp. 508–520, 2018, doi: 10.1016/S0140-6736(18)31129-2.
[2] J. Zeidan, E. Fombonne, M. Scorah, E. Ibrahim, A. Durkin, C. Saxena, and A. Elsabbagh, “Global prevalence of autism: A systematic review update,” Autism Research, vol. 15, no. 5, pp. 778–790, 2022, doi: 10.1002/aur.2696.
[3] K. A. Shaw, M. A. Bilder, L. A. McArthur, et al., “Prevalence and Early Identification of Autism Spectrum Disorder Among Children Aged 4 and 8 Years — Autism and Developmental Disabilities Monitoring Network, 16 Sites, United States, 2022,” MMWR Surveillance Summaries, vol. 74, no. 2, 2025.
[4] L. Franz, A. Chambers, A. von Isenburg, and C. de Vries, “Early intervention for autism spectrum disorder: A review of reviews,” Developmental Medicine & Child Neurology, 2022, doi: 10.1111/dmcn.15258.
[5] M. Sandbank, J. Bottema-Beutel, A. Crowley, et al., “Intervention effects on language in children with autism: A Project AIM meta-analysis,” BMJ, vol. 383, e076733, 2023, doi: 10.1136/bmj-2023-076733.
[6] A. J. O. Dede, W. Xiao, N. Vaci, M. X. Cohen, and E. Milne, “Exploring EEG resting state differences in autism: sparse findings from a large cohort,” Molecular Autism, 2025, doi: 10.1186/s13229-025-00647-3.
[7] W. S. Neo, D. Foti, B. Keehn, and B. Kelleher, “Resting-state EEG power differences in autism spectrum disorder: A systematic review and meta-analysis,” Translational Psychiatry, vol. 13, art. 389, 2023, doi: 10.1038/s41398-023-02681-2.
[8] R. Das, A. Arora, and R. A. A. Inamdar, “Machine learning approaches for autism spectrum disorder diagnosis: a systematic review,” Progress in Neuro-Psychopharmacology & Biological Psychiatry, vol. 122, 2023, doi: 10.1016/j.pnpbp.2022.110705.
[9] Y. Li, J. Xu, and X. Chen, “EEG-based autism spectrum disorder classification: A systematic review of machine learning and deep learning methods,” Computers in Biology and Medicine, vol. 173, 2024, art. 108075, doi: 10.1016/j.compbiomed.2024.108075.
[10] P. Soto-Icaza, P. Soto-Fernández, L. Kausel, et al., “Oscillatory activity underlying cognitive performance in children and adolescents with autism: a systematic review,” Frontiers in Human Neuroscience, vol. 18, art. 1320761, 2024, doi: 10.3389/fnhum.2024.1320761.
[11] L. Ronconi, A. Molteni, M. Casartelli, et al., “(Beta-band related) oscillatory dynamics and inhibitory control alterations in autism spectrum disorder,” NeuroImage: Clinical, 2020, doi: 10.1016/j.nicl.2020.102484.
[12] A. Moliadze, C. Brodski, and S. Caspers, “Significance of beta-band oscillations in autism spectrum disorders during motor response inhibition tasks,” Brain Topography, 2020, doi: 10.1007/s10548-020-00765-6.
[13] M. N. A. Tawhid, A. Al-Azzawi, M. Shouran, et al., “A spectrogram image based intelligent technique for autism spectrum disorder detection using EEG signals,” PLOS ONE, vol. 16, no. 6, e0253094, 2021, doi: 10.1371/journal.pone.0253094.
[14] M. Z. Ullah, S. Ahmad, and M. A. U. Khan, “Grid-tuned ensemble models for 2D spectrogram-based EEG autism classification,” Biomedical Signal Processing and Control, 2024.
[15] R. Djemal, H. Bourouba, and M. A. Alimohamad, “EEG-based autism diagnosis using wavelet, entropy, and neural networks,” BioMed Research International, 2017.
[16] M. Rosenblatt, L. Tejavibulya, R. Jiang, S. Noble, and D. Scheinost, “Data leakage inflates prediction performance in connectome-based machine learning models,” Nature Communications, vol. 15, art. 1829, 2024, doi: 10.1038/s41467-024-46150-w.
[17] M. N. A. Tawhid et al., “A spectrogram image based intelligent technique for automatic detection of autism spectrum disorder from EEG,” PLOS ONE, vol. 16, no. 6, e0253094, 2021, doi: 10.1371/journal.pone.0253094.
[18] F. Fahmi, M. Melinda, P. D. Purnamasari, E. Elizar, and A. Rafiki, “Recognition of EEG Features in Autism Disorder Using Stationary Wavelet Transform and Fisher Linear Discriminant Analysis,” Diagnostics, vol. 15, no. 18, art. 2291, 2025, doi: 10.3390/diagnostics15182291.
[19] R. Djemal, K. AlSharabi, S. Ibrahim, and A. Alsuwailem, “EEG-Based Computer Aided Diagnosis of Autism Spectrum Disorder Using Wavelet, Entropy, and ANN,” BioMed Research International, vol. 2017, art. 9816591, 2017, doi: 10.1155/2017/9816591.
[20] M. J. Alhaddad, “BCI Datasets at King AbdulAziz University (KAU),” accessed online, 2025.
[21] BCI2000.org, “Technical Reference: BCI2000 File Format,” updated 2025.
[22] P. Arpaia et al., “A Systematic Review of Techniques for Artifact Detection and Artifact Category Identification in Electroencephalography from Wearable Devices,” Sensors, vol. 25, no. 18, art. 5770, 2025, doi: 10.3390/s25185770.
[23] L. Pion-Tonachini, K. Kreutz-Delgado, and S. Makeig, “ICLabel: An automated electroencephalographic independent component classifier, dataset, and website,” NeuroImage, vol. 198, pp. 181–197, 2019, doi: 10.1016/j.neuroimage.2019.05.026.
[24] J. A. D. Delorme, A. M. Mullen, and S. Makeig, “EEGLAB, SIFT, ICLabel toolboxes for reproducible EEG preprocessing,” Frontiers in Neuroscience, 2019.
[25] O. Dimigen, “Optimizing the ICA-based removal of ocular EEG artifacts from free viewing experiments,” NeuroImage, vol. 207, art. 116117, 2020.
[26] M. Rosenblatt et al., “Data leakage inflates prediction performance in connectome-based machine learning models,” Nature Communications, vol. 15, art. 1829, 2024, doi: 10.1038/s41467-024-46150-w.
[27] N. M. Gosala and A. K. Amalanathan, “EEG signal feature extraction using different wavelet transforms and statistical techniques,” Biomedical Signal Processing and Control, vol. 85, art. 104811, 2023, doi: 10.1016/j.bspc.2023.104811.
[28] J. Li et al., “Identification of autism spectrum disorder based on electroencephalography: A systematic review,” Computers in Biology and Medicine, vol. 173, art. 108075, 2024, doi: 10.1016/j.compbiomed.2024.108075.
[29] A. Y. A. Amer, “Global-local least-squares support vector machine (GLocal-LS-SVM),” PLOS ONE, vol. 18, no. 4, e0285131, 2023, doi: 10.1371/journal.pone.0285131.
[30] R. Guido et al., “An overview on the advancements of support vector machine models in healthcare,” Information, vol. 15, no. 4, art. 235, 2024.
Copyright (c) 2026 melinda melinda, Muhammad Irhamsyah, Sri Rahayu Ade, Saifullah Nur Muhammad, Yunidar Yunidar, Nurlida Basir (Author)
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution-ShareAlikel 4.0 International (CC BY-SA 4.0) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
