Assessment of Fatigue in Robot-Mediated Upper Limb Interactions Using EEG

Abstract

Fatigue experienced during post-stroke rehabilitation and its implications for the therapy outcome are often overlooked in existing rehabilitation programmes. Past studies have shown that intensive and repetitive robot-mediated upper limb therapies improve the neuroplasticity of stroke survivors. However, it is more likely that the increased motor/cognitive processing demands required during post-stroke motor retraining exercises may exacerbate stroke patients’ fatigue levels. The elevated fatigue levels may impair motivation and compliance to effectively perform the rehabilitation tasks and the long-term commitment towards it. Hence, it is highly questionable whether continuing a stroke therapy while or beyond fatigue would impede motor performance and motor skill relearning during the session. While robot-mediated rehabilitation has been gaining traction over the past 40 years, EEG feature modulations associated with fatigue induced by robot-mediated interactions has not yet been comprehensively explored. Personalised rehabilitation sessions that incorporate the knowledge of patient fatigue levels and the effects of fatigue on brain activity are thought to improve the intervention’s efficiency. Moreover, EEG-based fatigue indices could also be used to mitigate fatigue accumulated during human-robot collaboration tasks, thereby managing the fatigue-related risks in the automotive industry. Therefore, the present research aims to investigate the modulations in spectral and nonlinear EEG features due to fatigue in a range of robot-mediated interactions involving gross motor, fine motor and visuomotor tracking tasks. This research work analysed and evaluated EEG correlations of fatigue induced by three different robot-mediated interactions using two experimental studies. A comparison of EEG spectral feature modulations following robot-mediated gross motor and fine motor interactions was conducted in experiment 1. An in-depth analysis of spectral and nonlinear EEG feature modulations during a robot-mediated visuomotor tracking task was conducted in experiment 2. Healthy participants were considered since this is an early-stage investigation. The two experiments have shown that fatigue was induced during robot-mediated interactions and has differently changed the EEG features and cortical sites depending on the type of interaction. Experiment 1 revealed that the robot-mediated gross motor interactions most likely change the EEG activity around the central and parietal brain regions. In contrast, this experiment found that the robot-mediated fine motor interactions most likely change the EEG activity around the frontopolar and central brain regions. Experiment 2 revealed that the robot-mediated visuomotor tracking tasks most likely change the EEG activity around the central, parietal and occipital brain regions. These observations were supported by the subjective measures of the level of fatigue. The correlation analysis performed in experiment 2 also revealed that the participants who maintained increased tracking accuracies during the robot-mediated visuomotor tracking task experienced an increase in their physical fatigue level, thereby contributing to a greater change in EEG features. Taken together, the findings presented in this thesis suggest that the modulations in EEG features and the cortical regions that are mostly affected due to fatigue induced by the robot-mediated interactions are specific to the physical and cognitive nature of the task performed. Therefore, the findings presented in this thesis confirm the hypothesis of this research “EEG correlates of fatigue during robot-mediated interactions are specific to the physical or cognitive nature of the task and the differences in the usage of proximal or distal upper limb”.