警示回饋對於腦波及行為上的影響及其於疲勞駕車之應用

Abstract

過去許多研究提出，當人們處於疲勞的狀態同時在精神上產生負面影響，伴隨著生理認知行為上發生錯誤，更可能是造成重大意外的關鍵因素，然而利用刺激提醒並結合行為及其伴隨的腦波變化的研究仍相當不足。本研究將有系統探討疲勞駕車在認知行為和腦波上的變化，並提出如何應用於真實生活中。本研究在虛擬實境的環境，模擬真實長時間駕車的場景，使受測者在虛擬場景中進行長時間駕車並維持注意力的實驗，目的在於探討在需要長時間維持注意的情況下，產生疲勞所造成的認知錯誤，並利用聲音訊號的刺激提醒之後，觀察腦波以及行為表現上的變化。本研究更進一步發展一個「基於腦波特徵的疲勞偵測及減緩系統」透過腦波即時監測以防止疲勞所造成的認知錯誤。 在實驗中，受測者的腦波長期與暫態的動態變化會透過獨立訊號分析、時域頻域轉換等方法進行分析比較。研究結果歸納出當受測者處於疲勞造成的打瞌睡狀態，枕葉腦區皮質的腦波在α和θ頻帶的能量會產生上升變化，行為的表現會下降。疲勞受測者在接收警示提醒之後，行為的表現會提升，同時枕葉腦波活動在α和θ頻段的上升的能量會下降接近清醒的狀態。從腦波的比較，警示的效果可以持續30秒或是更久的時間。由此，我們可以藉由腦波萃取出個人的認知清醒狀況資訊，更進一步的，將其應用在即時判斷監測認知上的變化。然而，本研究也提出，警示的效果可能會隨著實施次數的增加而減少，故本研究提出藉由機器學習的技術即時判斷警示效果，給予不同的刺激變化，來維持提醒受測者的警示效果。 本研究呈現的結果明確的指出藉由腦波的變化，可以判斷認知相關的資訊，更進一步利用「基於腦波監測的疲勞預測及減緩系統」來防止認知錯誤所可能引發的災禍事件。

Research has indicated that fatigue is a critical factor in cognitive and behavioral lapses because it negatively affects an individual’s internal state, which is then manifested physiologically. However, to our best knowledge, no study has assessed the EEG correlated of improved task performance following arousing signals. This study investigates brain dynamics and behavioral changes in response to arousing auditory signals presented to individuals experiencing momentary cognitive lapses, due to fatigue, during a sustained-attention task. Electroencephalographic (EEG) and behavioral data were simultaneously collected during virtual-reality (VR) based driving experiments, in which subjects were instructed to maintain their cruising position and compensate for randomly induced lane deviations using the steering wheel. This study further demonstrates the feasibility of an on-line closed-loop EEG-based fatigue prediction and mitigation system that detects physiological change and can thereby prevent fatigue-related cognitive lapses. Moreover, this work compares the efficacy of fatigue prediction and mitigation between the EEG-based and a non-EEG-based method. Each participant was instructed to maintain his/her cruising position at all times during the experiment. Each participant’s EEG signal was monitored continuously and a warning was delivered in real time to participants once the EEG signature of fatigue was detected. 30-channel EEG data were analyzed by independent component analysis and the short-time Fourier transform. Across subjects and sessions, intermittent performance during drowsiness was accompanied by characteristic spectral augmentation or suppression in the alpha- and theta-band spectra of a occipital component, corresponding to brief periods of normal (wakeful) and hypnagogic (sleeping) awareness and behavior. The improved behavioral performance was accompanied by concurrent spectral suppression in the theta- and alpha-bands of the occipital component. The effects of auditory feedback on spectral changes lasted 30 s or longer. The results of this study demonstrate the amount of cognitive state information that can be extracted from noninvasively recorded EEG data and the feasibility of online assessment and rectification of brain networks exhibiting characteristic dynamic patterns in response to momentary cognitive challenges. However, study results also showed reduced feedback efficacy (i.e., increased response times to lane deviations) accompanied by increased alpha-power due to the effects of habituation to repeat warnings. This study further proposes a feedback efficacy assessment system to accurately estimate the efficacy of arousing warning signals delivered to drowsy participants by monitoring the changes in their EEG power spectra immediately thereafter. The results of this study explore the amount of cognitive state information that can be extracted from noninvasively recorded EEG data and clearly demonstrate and validate the efficacy of this on-line closed-loop EEG-based fatigue prediction and mitigation mechanism to identify cognitive lapses that may lead to catastrophic incidents in countless operational environments.