Related Experiment Video
Updated: Aug 30, 2025
06:37
Electroencephalography Network Indices as Biomarkers of Upper Limb Impairment in Chronic Stroke
Published on: July 14, 2023
953
Improving EEG-Based Driver Distraction Classification Using Brain Connectivity Estimators
Dulan Perera1, Yu-Kai Wang2, Chin-Teng Lin2
1School of Science, Computing and Engineering Technologies, Swinburne University of Technology, Hawthorn, VIC 3122, Australia.
Sensors (Basel, Switzerland)
|August 26, 2022
Summary
Brain connectivity patterns detected via electroencephalogram (EEG) can classify driver distraction. Partial Directed Coherence (PDC) demonstrated the highest accuracy in distinguishing distracted from non-distracted driving states.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Human-Computer Interaction
Background:
- Driver distraction is a major cause of road accidents.
- Electroencephalogram (EEG) offers a non-invasive method to monitor brain activity.
- Brain connectivity analysis can reveal complex neural dynamics.
Purpose of the Study:
- To develop and evaluate an EEG-based method for classifying driver distraction.
- To assess the efficacy of various brain connectivity estimators as features for distraction detection.
- To identify the most accurate connectivity measure for distinguishing distracted driving states.
Main Methods:
- Ten participants performed driving tasks in a virtual reality environment under distracted and non-distracted conditions.
- Independent Component Analysis (ICA) was used to extract relevant brain activity.
- Granger-Geweke Causality (GGC), Directed Transfer Function (DTF), Partial Directed Coherence (PDC), and Generalized Partial Directed Coherence (GPDC) were employed as connectivity estimators.
- Support Vector Machine (SVM) with Radial Basis Function (RBF) kernel was utilized for classification.
Main Results:
- PDC achieved the highest classification accuracy (86.19%) for driver distraction.
- GGC (82.27%), GPDC (80.95%), and DTF (70.02%) also showed varying degrees of classification performance.
- Further analysis identified optimal window settings for PDC to differentiate driving states.
Conclusions:
- Brain connectivity estimators, particularly PDC, show significant potential for EEG-based driver distraction classification.
- This approach offers a promising avenue for developing advanced driver safety systems.
- Objective assessment of driver attention using neurophysiological signals is feasible.
