Three drowsiness categories assessment by electroencephalogram in driving simulator environment

Area of Science:

• Neuroscience
• Transportation Safety
• Biomedical Engineering

Background:

• Traffic accidents are a major global issue, with driver drowsiness being a primary cause.
• Drowsiness is difficult to prevent as it relates to physiological conditions.
• Electroencephalogram (EEG) is a common method for assessing drowsiness, as it originates in the central nervous system.

Purpose of the Study:

• To estimate driver drowsiness using frequency-domain and time-domain EEG analysis.
• To validate EEG findings with subjective (Karolinska Sleepiness Scale - KSS) and objective (Facial Expression Evaluation - FEE) measures.
• To determine the accuracy of EEG parameters in differentiating between alert, weakly drowsy, and strongly drowsy states.

Main Methods:

• Collected EEG data from subjects in a simulated driving environment.
• Performed frequency-domain and time-domain analysis on EEG signals.
• Utilized KSS and FEE to categorize subjects into three drowsiness levels: alert, weak drowsiness, and strong drowsiness.

Main Results:

• Six EEG parameters (absolute/relative alpha power, β/α ratio, (θ+α)/β ratio, Hjorth activity, Hjorth mobility) showed significant differences across the three drowsiness states (P < 0.001).
• Combined EEG parameters with KSS and FEE achieved high accuracy (up to 92.9%) in detecting drowsiness.
• Statistically significant differences were observed in EEG parameters between alert, weak, and strong drowsiness conditions.

Conclusions:

• EEG parameters are effective in detecting varying levels of driver drowsiness.
• The proposed EEG-based method, validated by KSS and FEE, offers a promising approach for real-time drowsiness detection in driving.
• This research contributes to enhancing road safety by providing a reliable method for monitoring driver alertness.