Copertina

Real-time and energy-efficient attention monitoring using single-channel EEG for everyday life

Attention monitoring is essential in tasks requiring sustained cognitive focus. Electroencephalography (EEG) offers a portable, cost-effective, and non-invasive solution; however, its signal complexity requires advanced processing. This study aims to develop an accurate and efficient method for clas...

Descrizione completa

Salvato in:
Dettagli Bibliografici
Autori principali: Jahanjoo, Anice (Autore) , Bauer, Vimala (Autore) , Khooyooz, Soheil (Autore) , Haghi, Mostafa (Autore) , Taherinejad, Nima (Autore)
Natura: Article (Journal)
Lingua:inglese
Pubblicazione: 21 January 2026
In: IEEE access
Year: 2026, Volume: 14, Pages: 12396-12406
ISSN:2169-3536
DOI:10.1109/ACCESS.2026.3656851
Accesso online:Verlag, kostenfrei, Volltext: https://doi.org/10.1109/ACCESS.2026.3656851
Testo
Note sull'autore:Anice Jahanjoo, Vimala Bauer, Soheil Khooyooz, Mostafa Haghi, and Nima Taherinejad
Descrizione
Riassunto:Attention monitoring is essential in tasks requiring sustained cognitive focus. Electroencephalography (EEG) offers a portable, cost-effective, and non-invasive solution; however, its signal complexity requires advanced processing. This study aims to develop an accurate and efficient method for classifying attention using a minimal EEG setup suitable for wearable applications. We propose an optimized single-channel EEG approach that combines Recursive Feature Elimination (RFE) for feature selection with a customized eXtreme Gradient Boosting (cXGB) classifier. The use of a single EEG electrode enables compatibility with commercially available, user-friendly EEG headbands, making real-time attention monitoring feasible in daily life. Our findings show that the proposed cXGB model achieves 98.29% binary and 94.25% three-class classification accuracy under subject-wise cross-validation, outperforming previous multi-channel approaches by 3%. Furthermore, the optimized feature set reduces processing time, memory usage, and energy consumption on both high-end (Intel i9) and low-power (Raspberry Pi) devices. This work provides an efficient and scalable solution for real-time, individualized attention monitoring on wearable platforms.
Descrizione del documento:Gesehen am 02.06.2026
Online veröffentlicht: 21. Januar 2026, Artikelversion: 26. Januar 2026
Descrizione fisica:Online Resource
ISSN:2169-3536
DOI:10.1109/ACCESS.2026.3656851

Documenti analoghi