DOI: 10.12688/f1000research.174232.2 ISSN: 2046-1402

EEG-Based Discrimination of Autism Spectrum Disorder in Youth Ages 10–20 Using a Wearable Headband: A Pilot Study of Temporal Beta and Gamma Power

Omayya Murad, Mohammad Malkawi, Methaq Khadum, Nour Faouel, Khalid Batterjee, Samer Dashi, Hala Leberrara, Ahmad Arafat
Show PDF Cite
Background Autism Spectrum Disorder (ASD) diagnosis relies on behavioral assessment, which can be time-consuming and subjective. This proof-of-concept study asks if minimal EEG band measurements can differentiate youth with ASD from neurotypical peers, a step toward possible screening approaches; this small, single-site sample cannot itself establish screening utility. Methods Twelve participants aged 10–20 (5 controls, 7 ASD) were recorded with a four-sensor Muse 2 headband covering frontal and temporal regions. Five EEG bands — alpha, beta, gamma, delta, theta — were measured and analyzed per participant. No offline artifact-rejection pipeline (e.g., Independent Component Analysis) was applied beyond the device’s onboard preprocessing; this is treated as a central limitation, addressed explicitly in the Discussion and Limitations. Results Absolute gamma power at the right temporal sensor (TP10) and absolute beta power at the left temporal sensor (TP9) showed the largest group differences and the strongest cross-validated classification performance. Because gamma is highly susceptible to electromyographic (EMG) and motion contamination, and no validated artifact-rejection pipeline was applied, these findings are reported as candidate physiological screening features rather than confirmed neural correlates of ASD; movement or facial-muscle activity during unconstrained recording cannot be excluded as a contributor. Individual-level values for both features, overlaid on group summary statistics, are shown in Figure 5. Conclusion Beta at TP9 and gamma at TP10 emerge as candidate physiological features warranting further study, whether the underlying signal is purely neural or partly reflects physiological/behavioral activity such as muscle tension. Given the small sample (n = 12), the absence of validated artifact rejection, single-site recruitment, and no independent-sample testing, this study does not establish clinical or educational screening value, and no such application should be inferred here. Confirming utility will require replication in larger, multi-site, demographically broader samples, with simultaneous EMG/EOG recording, validated artifact-rejection procedures, and prospective testing in independent participants.

More from our Archive