Automatic Detection of Driver Fatigue Based on EEG Signals Using a Developed Deep Neural Network

Sheykhivand, Sobhan; Rezaii, Tohid Yousefi; Mousavi, Zohreh; Meshgini, Saeed; Makouei, Somayeh; Farzamnia, Ali; Danishvar, Sebelan; Kin, Kenneth Teo Tze · 2022 · OpenAlex-citations

DOI: 10.3390/electronics11142169

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 fatigue, a leading cause of road accidents resulting in significant global mortality and economic loss. While physiological signals, particularly electroencephalogram (EEG) data, offer precise non-invasive monitoring, existing detection methods often rely on manual feature extraction, which increases computational complexity and lacks generalizability. Furthermore, prior research has largely neglected mental fatigue induced by monotonous driving conditions and failed to account for environmental noise, such as engine sounds. To address these gaps, the authors propose an automatic driver fatigue detection system that utilizes a deep neural network to analyze raw EEG data in the presence of environmental noise. The researchers collected EEG data from 11 graduate students using a driving simulator designed to induce mental fatigue through a uniform, traffic-free highway route. Data were recorded using a 32-channel EEG recorder at 1000 Hz, with participants driving for 60–100 minutes or until fatigue was confirmed via performance metrics and standardized questionnaires (Chalder and Lee scales). The dataset, comprising 5500 samples, was prepared according to established criteria. The proposed method employs a hybrid deep convolutional neural network–long short-term memory (CNN–LSTM) architecture. This model hierarchically extracts features directly from raw EEG signals corresponding to six active brain regions identified through independent component analysis, eliminating the need for manual feature selection. The results demonstrate that the CNN–LSTM network effectively learns features from raw EEG data and achieves higher precision rates than previous comparative approaches for two-stage driver fatigue categorization. The system successfully distinguishes between normal and fatigued states despite the presence of environmental noise. By focusing on specific active regions rather than all channels, the method reduces computational complexity while maintaining high accuracy. The study confirms that deep learning models can serve as robust, end-to-end solutions for fatigue detection, outperforming traditional methods that rely on hand-crafted features like entropy measures or spectral analysis. The significance of this work lies in its contribution to automated road safety systems. By providing a comprehensive dataset and a novel algorithm that accounts for real-world variables like environmental noise and mental fatigue, the research offers a practical foundation for developing reliable driver monitoring systems. The proposed approach’s high precision and speed make it suitable for integration into vehicles to warn drivers during drowsiness, potentially reducing accident rates. This study advances the field by shifting from manual feature engineering to automated hierarchical feature learning, addressing key limitations in prior EEG-based fatigue detection research.

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.

Topics

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

• drowsiness detection algorithms

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