A multimodal visual fatigue assessment model based on back propagation neural network and XGBoost

An experiment was conducted using a subjective questionnaire, ophthalmological parameters, electroencephalogram (EEG) signals, electrocardiogram (ECG) signals, and eye-tracking parameters. The goal was to investigate the impact of display modes (2D, normal 3D, and enhanced 3D) on visual fatigue. The results of paired samples t-tests for both subjective and objective parameters indicated a significant influence of the display mode on visual fatigue. A visual fatigue assessment model employing multimodal parameters and based on a backpropagation neural network (BPNN) and XGBoost was proposed. The results suggested that the model was effective in predicting visual fatigue states, which the F1 and Accuracy value were both 0.94.