Multivariate Multiscale Entropy: An Approach to Estimating Vigilance of Driver

Ahammed, Kawser; Uddin Ahmed, Mosabber · 2023 · Crossref

DOI: 10.4108/eetpht.8.3432

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 gap in driver vigilance estimation techniques that fail to analyze physiological signals within the complexity domain. While existing methods rely on time-domain analysis of video, multi-sensor, or physiological data, this research introduces the multivariate multiscale entropy (MMSE) method to characterize driver cognitive states. The primary motivation is to provide a more robust indicator for vigilance by quantifying the complexity of brain and eye movement signals, thereby improving the detection of fatigue and drowsiness in driving scenarios. The researchers utilized a dataset comprising electroencephalogram (EEG) signals from posterior and temporal brain sites, forehead EEG, and electrooculogram (EOG) signals, sampled at 1000 Hz and downsampled to 200 Hz. Data preprocessing involved band-pass filtering (1–75 Hz) and the extraction of differential entropy (DE) features across five frequency bands (delta, theta, alpha, beta, gamma) and a high-resolution 2 Hz spectrum. The MMSE algorithm was applied to these features and percentage of eye closure (PERCLOS) values to evaluate complexity across multiple time scales. Driver states were categorized into awake, tired, and drowsy based on PERCLOS thresholds. A support vector machine (SVM) classifier, trained with 5-fold cross-validation, was employed to discriminate between these cognitive states using the derived MMSE features. The results demonstrated statistically significant differences (p < 0.01) in complexity profiles among forehead EEG, brain EEG, and EOG signals, with forehead EEG exhibiting the highest complexity. Analysis of cognitive states revealed that the awake state possessed significantly higher multivariate sample entropy values (1.0828 ± 0.4664) compared to the tired (0.7841 ± 0.3183) and drowsy (0.2938 ± 0.1664) states. This indicates that awake drivers exhibit more complex, less regular signal patterns, whereas fatigue leads to more regular, lower-complexity signals. The SVM classifier achieved a classification accuracy of 76.2%, with the drowsy state showing 100% sensitivity and the awake state demonstrating 92.9% specificity. Statistical tests, including ANOVA and t-tests, confirmed significant distinctions between all cognitive state pairs. The study concludes that MMSE provides a viable framework for estimating driver vigilance in the complexity domain. The findings suggest that complexity profiles, particularly from forehead EEG, serve as effective indicators for distinguishing between alert and fatigued states. Although the current accuracy is promising, it is lower than some conventional methods, suggesting potential for optimization through neurofeedback integration or advanced algorithm tuning. The authors note that while the method is effective for monotonous highway driving, its applicability to high-mental-load urban driving requires further investigation. This work establishes a foundation for developing real-time vigilance monitoring systems based on complexity science.

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
• drowsiness
• microsleep
• neuro workload indices

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: validation psychometrics, tool software