Real Time Vigilance Detection using Frontal EEG

Ganesh, Siddarth; Gurumoorthy, Ram · 2021 · Crossref

DOI: 10.5121/ijcsit.2021.13104

archive: archived pipeline: cataloged verified

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

Summary

This paper addresses the critical safety and performance issues caused by vigilance drops in monotonous tasks, such as driving, manufacturing, and remote learning. The authors argue that existing solutions, including computer vision and peripheral physiological sensors like photoplethysmography, are either intrusive, require complex infrastructure, or rely on lagging indicators. To overcome these limitations, the study proposes a non-invasive, real-time vigilance detection system using frontal electroencephalogram (EEG) data. The core hypothesis is that changes in the power spectrum of the frontal theta band (4–8 Hz) serve as an early and accurate predictor of an individual’s attention level. The methodology involved four participants aged 14 to 50 who wore a commercial wireless EEG headband (Muse-S) connected to a mobile application via Bluetooth Low Energy. Data was collected from a single frontal electrode (AF7) at a 256 Hz sampling rate. To ensure robustness across circadian rhythms, sessions were conducted at three distinct times: morning, post-lunch, and late evening. Each session lasted three minutes, including a baseline period. Two protocols were tested: an "instructed" mode where participants opened and closed their eyes on command, and a "natural" mode where eye status was self-tagged. Data processing involved calculating the power spectral density using Discrete Fourier Transform every five seconds. A personalized vigilance threshold was established based on the mean theta band power during the initial 30-second baseline (eyes closed). Subsequent epochs were classified as vigilant or non-vigilant based on whether the theta power exceeded this threshold, scaled by a factor of 1.1. The results demonstrated that frontal theta band power effectively distinguishes between vigilant (eyes open) and non-vigilant (eyes closed) states. The system achieved an average prediction accuracy of 89.30% across all sessions. Specifically, the natural protocol yielded a higher average accuracy of 90.91% compared to 87.70% for the instructed protocol. Statistical analysis via a T-test indicated no significant difference between the two methods (p-value = 0.098). Confusion matrix analysis revealed that the instructed method had a higher error rate in predicting open eyes when they were actually closed (23.21%), likely due to the delay in cognitive engagement after the instruction. In contrast, the natural method reduced this error to 11.62%, supporting the hypothesis that natural transitions provide cleaner data for vigilance assessment. The study concludes that a single-channel frontal EEG system offers a practical, low-cost, and non-invasive solution for real-time vigilance monitoring. By detecting central nervous system changes earlier than peripheral or behavioral indicators, this approach can be integrated into various applications, such as driver drowsiness alerts, industrial safety monitoring, and e-learning attention tracking. The authors suggest that future work should expand the participant pool and explore machine learning models to further enhance prediction accuracy and adaptability across diverse environments.

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
• neuro workload indices
• vigilance
• microsleep
• drowsiness

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
• Methodological Resource: tool software