| dc.description.abstract | Mental fatigue is described as a feeling experienced by some individuals either during or after prolonged cognitive activities. This sensation is very common in modern daily life and generally includes fatigue, reduced moti-vation and interest in the current activity, and decreased commitment to tasks. In this study, 50 participants (25 females and 25 males) aged 18 to 30 years (mean age: 24 years) participated. After obtaining informed con-sent, all participants were evaluated for quantitative electroencephalog-raphy (EEG) characteristics. To induce fatigue, participants were asked to complete the AX-CPT task for 90 minutes. Then, half of the participants were exposed to transcranial photobiomodulation therapy with a wave-length of 850 nm, a power density of 60 mW/cm², and an energy density of 4 J/cm². After light exposure, participants were reassessed for EEG characteristics to investigate the effects of transcranial photobiomodula-tion on brainwaves after inducing mental fatigue. Data preprocessing was performed to enhance signal quality and improve the accuracy and preci-sion of the classification model. The data were randomly augmented dur-ing training to increase the dataset and enable the model to classify new instances more accurately. In this study, three pre-trained neural net-works—LSTM, GCN, and ResGCN—were used to classify the signals. Ultimately, the ResGCN network demonstrated superior performance, achieving the lowest loss rate during learning (0.1751), the highest accura-cy in the final epoch (95.38%), and the highest classification accuracy on test data (91.67%), making it the optimal network.
Keywords:
Mental fatigue, transcranial photobiomodulation, neural networks, ma-chine learning, deep learning. | en_US |
