Mental Work Load Assessment Using Brain Connectivity Based Features of Simultaneous EEG and fNIRS

Mental Work Load (MWL) is essential for evaluating human performance, reflecting brain activity and information processing during tasks. Accurate MWL assessment is crucial for precise task execution and is widely applicable in system design and evaluation for vehicle control, human-computer interfaces, brain-computer interfaces, and educational fields. This study assesses MWL using simultaneous EEG and fNIRS signals. EEG provides high temporal resolution but is noise-prone, while fNIRS is robust against noise but has low temporal resolution, making them complementary. MWL levels were classified using Functional Brain Connectivity (FBC) features extracted from EEG and fNIRS signals and statistical-based features from the differential of oxygenated and deoxygenated hemoglobin (HbD). Our results indicate that using functional connectivity features, the classification accuracy for low and high MWL levels was $\mathbf{7 0. 9} \%$ for EEG, $\mathbf{6 5. 5} \%$, and $\mathbf{6 5. 9} \%$ for OXY and DEOXY signals, respectively, and $73.3 \%$ when combining EEG and fNIRS features. In the hybrid classification by PLI feature extraction, 8 to $12 \%$ accuracy has been improved compared to the classification using EEG and fNIRS alone. These findings highlight the effectiveness of combining EEG and fNIRS for MWL assessment and have significant implications for various practical applications.