Heart Rate Variability-Based Driver Drowsiness Detection and Its Validation With EEG

Fujiwara, Koichi; Abe, Erika; Kamata, Keisuke; Nakayama, Chikao; Suzuki, Yoko; Yamakawa, Toshitaka; Hiraoka, Toshihiro; Kano, Manabu; Sumi, Yukiyoshi; Masuda, Fumi; Matsuo, Masahiro; Kadotani, Hiroshi · 2018 · OpenAlex-citations

DOI: 10.1109/tbme.2018.2879346

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 drowsy driving, which significantly increases the risk of traffic accidents. While electroencephalography (EEG) is the standard for sleep scoring, it is impractical for real-time driver monitoring due to motion artifacts and physical restrictions. The authors propose a non-invasive, wearable-friendly drowsiness detection algorithm based on heart rate variability (HRV) analysis, validated against EEG-based sleep scoring. The method leverages the physiological link between sleep onset and autonomic nervous system changes reflected in cardiac signals. The methodology employs multivariate statistical process control (MSPC), an anomaly detection technique originally developed for epileptic seizure prediction. The algorithm monitors eight linear HRV features—five time-domain metrics (MeanNN, SDNN, RMSSD, Total Power, NN50) and three frequency-domain metrics (LF, HF, LF/HF)—extracted from RR interval data using a three-minute moving window. Because drowsy data is scarce, the system is trained exclusively on awake data to establish individual control limits for $T^2$ and $Q$ statistics. Drowsiness is flagged when these statistics exceed their limits continuously for more than ten seconds, reducing false positives caused by signal artifacts. Experiments were conducted using a driving simulator with 34 healthy participants who drove monotonous highway loops for 1.5 hours each. EEG and RR interval data were collected simultaneously. A sleep specialist identified sleep onsets (N1 stage) via EEG, defining the 15 minutes prior as "pre-N1" episodes. The algorithm was trained on 34 awake episodes and validated on 57 awake and 13 pre-N1 episodes. The $Q$ statistic demonstrated superior performance, detecting drowsiness in 12 out of 13 pre-N1 episodes (92% sensitivity), whereas the $T^2$ statistic detected only 8 (62% sensitivity). The mean detection time for the $Q$ statistic was approximately 642 seconds before sleep onset. The false positive rate for the $Q$ statistic was 1.7 per hour. Analysis suggested that some false positives correlated with awake $\alpha$ waves and eye blinking, potentially indicating microsleep. The study concludes that HRV-based anomaly detection is a viable, less intrusive alternative to EEG for monitoring driver drowsiness. By utilizing a simple linear method with only eight features, the approach is computationally efficient and suitable for wearable devices. The high sensitivity of the $Q$ statistic in detecting physiological changes prior to sleep onset supports the potential for developing practical driver-assistance systems that can warn drivers before they fall asleep, thereby enhancing road safety.

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-prismaquant on 2026-06-26; verification: verified_with_issues.

Topics

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

• drowsiness detection algorithms
• drowsiness
• microsleep
• workload measurement

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