Self-efficacy Measurement Method Using Regression Models with Anticipatory Gaze for Supporting Rehabilitation

Self-efficacy (SE) is important for task motivation in rehabilitation patients. We propose a method for measuring task-specific SE (SSE) using anticipatory gaze. Experiments were conducted using on-screen tasks with multiple difficulty levels to collect gaze and manipulation data. The four anticipatory gaze-related features and the subjective SSE were compared, and the highest correlation coefficient between the gaze feature and SSE was 0.573. We also constructed a regression model using machine-learning methods and verified results. The mean absolute error of the model was 11.5, suggesting the possibility of measuring SSE using gaze information. We plan to apply this method to real-world rehabilitation.