Toward Wearable EEG-based Alertness Detection System Using SVM with Optimal Minimum Channels

Yang, Mihong; Li, Huiyan; Sun, Xiaozhou; Yang, Li; Duan, Hailong; Che, Yanqiu; Han, Chunxiao · 2018 · Crossref

DOI: 10.1051/matecconf/201821403009

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the critical safety issue of driver drowsiness, a leading cause of high-fatality traffic accidents. The authors aim to develop a wearable electroencephalogram (EEG)-based alertness detection system that balances accuracy with the practical constraints of wearable technology, specifically the need for minimal electrode channels to ensure long-term comfort and convenience. While previous methods relied on subjective scales, video monitoring, or high-channel-count EEG setups, this research focuses on optimizing a low-channel EEG configuration for real-time, wearable application. To achieve this, the researchers constructed a simulated driving environment using Unity3D, featuring a highway scene where the vehicle randomly drifted from its lane. A single male subject (age 36) participated in a one-hour driving experiment during the early afternoon, a period associated with increased drowsiness. EEG signals were recorded from ten channels located in the frontal (FP1, FP2) and occipital (PO5, PO3, POZ, PO4, PO6, O1, OZ, O2) regions using a NeuroScan system. Data segmentation was performed based on the subject’s driving reaction time—the interval between vehicle drift and corrective steering. Segments with reaction times exceeding 2.5 seconds were classified as drowsy, while those under 1.5 seconds were classified as alert. Features extracted from these segments included variance and peak values, processed using a 5-second sliding window with 3-second overlap. A Support Vector Machine (SVM) classifier was employed to distinguish between alert and drowsy states. The results demonstrated that alertness could be classified with high efficiency using a minimal number of channels. Using all ten channels yielded an accuracy of 95.34%. However, specific combinations of fewer channels performed comparably or better. The single channel PO6 achieved 93.52% accuracy. The two-channel combination of FP1 and PO6 reached 95.85% accuracy. The highest accuracy of 96.11% was achieved with three-channel combinations, specifically FP1+PO6+PO5 and FP1+PO6+POZ. Adding more channels beyond this optimal set did not improve accuracy and sometimes reduced it. The authors identified FP1, PO5, and PO6 as the optimal minimal channel combination, considering both performance and the symmetry required for wearable cap design. The significance of this work lies in its demonstration that high-accuracy alertness detection is feasible with a minimal EEG channel setup, facilitating the development of comfortable, wearable monitoring systems for drivers. By identifying specific frontal and occipital channels as sufficient for robust classification, the study provides a practical pathway for integrating real-time physiological monitoring into automotive safety systems without the burden of complex, multi-electrode headgear. Future work will involve expanding feature sets and analyzing the relationship between EEG data and driving behavior in larger subject groups.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data