Classifying Features of Electroencephalography Signal to Detect Driver Drowsiness in the Early Drowsy Stage
DOI: 10.18502/jss.v6i(1-2).9288
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 severe accidents, by proposing an electroencephalography (EEG)-based detection method focused on the early stages of fatigue. The research was motivated by the limitations of existing detection methods, which often rely on elapsed time or multiple physiological sensors that are intrusive or dependent on environmental conditions. The authors aimed to determine if EEG signals could reliably detect drowsiness before it becomes extreme, specifically utilizing the Observer Rating of Drowsiness (ORD) to assess drowsiness levels rather than relying on time-based metrics. The experimental design involved 19 healthy male participants who underwent driving tests in a simulator. After excluding two subjects based on Maintenance of Wakefulness Test results, data from 17 subjects were analyzed. EEG signals were recorded from six channels (C3, C4, P3, P4, O1, O2) and processed using discrete wavelet transform (DWT) and independent component analysis (ICA) to remove artifacts and decompose signals into frequency bands. Twelve statistical and spectral features were extracted, including mean, standard deviation, kurtosis, energy, entropy, and power in various frequency bands. Neighborhood Component Analysis (NCA) was employed for feature selection, identifying six key features that significantly changed during the transition from alertness to early drowsiness (ORD level 2.5). These features were then input into four classifiers: k-nearest neighbor (KNN), support vector machine (SVM), classification tree, and Naive Bayes. The results demonstrated that the classification tree algorithm achieved the highest performance. When using data from all six EEG channels, the classification tree detected early drowsiness with an accuracy of 88.55%. However, the study found that using only the single-channel P4 data yielded superior performance, achieving a classification accuracy of 93.13% with the classification tree. This single-channel approach outperformed the multi-channel method and other classifiers, such as KNN and SVM, which showed lower average accuracies. The selected features, particularly the power of the alpha band in the 11–15 Hz range (alpha spindles), were identified as crucial indicators of the transition to drowsiness. The significance of this study lies in its demonstration that driver drowsiness can be effectively detected in its early stages using a minimal, single-channel EEG setup. The high accuracy achieved with the P4 channel suggests that complex multi-channel setups are unnecessary for this application, offering a more practical, cost-effective, and less intrusive solution for real-world implementation. By focusing on early detection, the proposed method provides sufficient time for intervention, potentially preventing the progression to extreme drowsiness and associated fatal accidents.
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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-06 |
| archive | success | canonical_url | — | — | 13 | 2026-06-09 |
| extract | success | cached | — | — | 2 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-07 |
| chunk | success | chunk | — | — | 1 | 2026-06-07 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-07 |
| promote | success | — | — | — | 1 | 2026-06-06 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | partial | — | — | — | 1 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified_with_issues.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
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