The Research Of Exercise-induced Fatigue Based On EEG Signals

Life is closely bound up with fatigue.Excessive psychological pressure,long-time working,big labor intensity can make someone exhausted,impact cognitive and motor function,even cause sudden diseases or accelerate existing disease progression.Particularly,exercise-induced fatigue can be easily caused by patients with motor dysfunction during rehabilitation treatment because of their anxiety and overexertion,which affects the effects of rehabilitation therapy or even aggravates their condition.Therefore,how to make an effective detection and evaluation of exercise-induced fatigue has become an intersecting research hotspot in the fields of neuroscience and rehabilitation engineering.Some research shows that with the development of exercise-induced fatigue,the neuronal activities in cerebral motor cortex are inhibited and the intensity of the brain neural network connectivity decreases,which provides a theoretical basis for the detection of exercise-induced fatigue through brain activity.The study of exercise-induced fatigue detection based on electroencephalogram(EEG)is explored in this thesis.The main work and achievements are as follows:(1)An experiment simulating rehabilitation training of upper extremity dysfunction is completed together by our laboratory and the University of Massachusetts Lowell.In this experiment,the EEG data of 20 subjects are collected and evaluated fatigue levels by the Borg CR10 scale.(2)In time and frequency domain,the information interaction between C3 and C4 channels in motor cortex are discussed by the New Causality method(NC)and Granger Causality method(GC)to reveal the mechanism of exercise-induced fatigue with the Borg CR10 scale.It is shown that with the development of exercise-induced fatigue: 1)the causal values between channel C3 and C4 decrease,which indicates the interaction of the causal information between C3 and C4 is weakened and validates the previous result that fatigue makes an inhibitory effect on motor neurons and impairs brain neural network connectivity;2)In frequency domain,the weakening trend of causality on Beta band is more obvious than its behavior on Alpha band,which validates the previous result that Beta waves has a closer relationship with exercise-induced fatigue than Alpha waves;3)Compared NC results and GC results,there is a significant linear relationship between NC values and Borg values in both time and frequency domain.But GC values are without this relationship,which indicates that the NC method can reveal the relationship between exercise-induced fatigue and the cerebral motor function regions accurately,and explain the exercise-induced fatigue mechanism.(3)Phase synchronization of multichannel in motor cortex is applied to the exercise-induced fatigue detection.The exercise-induced fatigue detection is classified and detected based on PLV by support vector machine,logistic regression and random forest algorithm,and the Phase Locked Value(PLV)is calculated as the feature of fatigue detection on Alpha and Beta waves respectively.The results show that the accuracy of fatigue detection based on phase synchronization are all above 69% for both Alpha and Beta waves,and the average fatigue detection accuracy of Beta waves is higher than Alpha waves by three methods,especially the average accuracy of exercise-induced fatigue detection by support vector machine for Beta waves is up to 78%.In summary,this thesis demonstrates clearly that the exercise-induced fatigue is bound up with the status of the cerebral motor function regions,and it is entirely possible to detect and evaluate exercise-induced fatigue based on single signal—EEG,which has certain reference value for the development of fatigue detection and evaluation system and promotes the development of intelligent rehabilitation therapy robots.