Driver Distraction Detection Using Bidirectional Long Short-Term Network Based on Multiscale Entropy of EEG

Zuo, Xin; Zhang, Chi; Cong, Fengyu; Zhao, Jian; Hamalainen, Timo · 2022 · Crossref

DOI: 10.1109/tits.2022.3159602

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 challenge of detecting driver distraction in realistic driving scenarios, a critical issue given that distraction is a major contributory factor in traffic accidents. While previous research has relied on vehicle behavioral features or controlled laboratory settings, the authors note that real-world distraction involves complex, multi-modal interactions that are difficult to capture with single-sensor approaches. The paper proposes a novel framework that combines Electroencephalography (EEG) signals with vehicle behavioral data to detect distraction more accurately. Specifically, it utilizes Multi-scale Entropy (MSE) to extract complexity features from EEG signals and employs a Bidirectional Long Short-Term Memory (BiLSTM) network to model temporal dependencies in the data. The experimental design involved six right-handed subjects performing driving trials on a real straight road. Data were collected using a wireless EEG headband (sampling at 256 Hz, focusing on the O1 and O2 occipital channels) and vehicle sensors recording speed and deceleration (50 Hz). Participants completed one normal driving trial and five distracted trials, where they were required to use a smartphone (WeChat) for at least three seconds. The methodology involved preprocessing the EEG signals using wavelet decomposition to isolate the alpha frequency band (8–13 Hz), which is correlated with distraction, and removing artifacts via thresholding. MSE was then calculated using a sliding window to quantify signal complexity across multiple time scales. Vehicle behavioral data were statistically analyzed to identify deviations from normal driving states. Finally, a BiLSTM classifier was trained to detect distraction using MSE features, traditional EEG features, and vehicle behavioral features. The results demonstrated that MSE values notably decreased after the onset of distraction, indicating a reduction in EEG signal complexity. Statistical analysis confirmed that driving performance, specifically speed and deceleration, significantly deviated from normal states during distracted trials. The BiLSTM model utilizing MSE features outperformed other entropy-based methods and classifiers relying solely on behavioral features. Furthermore, combining MSE features with vehicle behavioral features improved detection accuracy by 3% compared to using behavioral features alone. The study concludes that MSE effectively captures the dynamic changes in brain activity associated with distraction, and the proposed multi-modal BiLSTM framework offers a robust solution for real-time driver distraction detection in naturalistic driving 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.

• distraction detection algorithms
• drowsiness detection algorithms
• visual
• mind wandering

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, behavioral performance data
• Theoretical Contribution: conceptual framework