EEG-based driving intuition and collision anticipation using joint temporal-frequency multi-layer dynamic brain...

Area of Science:

• Neuroscience
• Cognitive Science
• Traffic Safety

Background:

• Intuition is vital for driving decisions and traffic safety.
• Understanding the neural basis of driving intuition is crucial for developing advanced safety systems.

Purpose of the Study:

• To investigate changes in brain activity during driving intuition training using advanced network analysis.
• To identify EEG biomarkers for predicting imminent driving collisions.

Main Methods:

• Utilized a novel "Joint Temporal-Frequency Multi-layer Dynamic Brain Network" (JTF-MDBN) analysis on EEG data.
• Compared initial and advanced phases of driving intuition training using multi-layer and single-layer network metrics.
• Analyzed EEG data in theta, alpha, and beta bands during alert and non-alert driving states, focusing on pre-collision time windows.

Main Results:

• Advanced intuition training led to more stable brain region activity.
• JTF-MDBN showed stronger connection strength in the alert state.
• Multi-layer analysis revealed higher modularity in the non-alert state (alpha and beta bands).
• Identified significant EEG features in the W4 window (1 second before collision) differentiating imminent collisions.
• Brain network biomarkers achieved 87.5% accuracy in predicting collisions using a linear kernel SVM.

Conclusions:

• Driving intuition training enhances neural network stability and connectivity.
• EEG-based brain network biomarkers can effectively predict driving collision risk.
• Findings support the development of brain-computer interface systems for intelligent driving hazard perception.