- 學位論文
探討長時間電生理訊號及心理量表記錄與研究生情緒與持續專注力之關聯性
Longitudinal electrophysiological and psychometric correlates of emotional state and sustained attention ability in graduate students
摘要
當今高等教育被視為個人發展中一個必要的階段；過去五十年來，全球中學畢業生之大學錄取率自10%增加至38%，已開發國家更高達60%以上，意即年輕成人族群進入大專院校就讀佔了絕大多數。數個橫向調查型研究顯示，學生為易受到壓力、憂鬱及心理疲勞負面影響之高風險族群，而這些因子常常因偶發性或常規性的睡眠不足而出現。本論文實現了一個多面向研究，於真實上課場景中針對研究生 (n=26) 探討情緒狀態及注意力導向課堂活動之關聯性。在實驗中的第一階段，受試者須使用Daily Sampling System手機應用程式回報每天夜晚的睡眠品質及長度、白天時想睡覺的程度與自評幸福感。心理統計資料分析的結果呈現主觀疲勞與壓力、日間睡意、睡眠長度和清醒時警覺程度具顯著的時間關聯性，同時具高度個體間變異性。第二階段中，受測者每兩週進行一次靜息態腦電圖與心電圖紀錄，並填寫抑鬱、焦慮和壓力量表。腦電圖頻譜分析發現，隨著不同情緒狀態，腦電圖也呈現顯著的變化；尤其是當壓力與焦慮程度升高，低頻的腦波活動也跟著增強。紀錄生理電訊號之後，受測者會於日常課堂參與一個特別設計的持續視覺專注力實驗。藉由排序出實驗中反應時間快慢的腦電圖頻譜顯示出非事件相關的腦波活動特徵，與實驗中隨機出現之視覺刺激帶來的視覺警覺程度具關連性。最後，實驗中收錄之腦波特徵可用來訓練機器學習之分類系統，經由受試者各自的腦波特徵校正後，開發能夠辨識不專注及情緒激發程度之腦機介面系統。
並列摘要
Higher education is now considered a necessary stage in personal development: over the past 50 years, the tertiary enrollment rate of secondary school graduates increased from 10% to 38% worldwide, exceeding 60% in developed countries — therefore making college and university students accounting for a significant portion of the young adult population. A number of cross-sectional studies report students as a high-risk group for experiencing adverse effects of stress, depression, and mental fatigue, which often coexist with occasional or regular sleep deprivation. This work processes the dataset longitudinally collected in a group of graduate school students (n=26) for investigating physiological correlates of emotional state and attention-driven classroom activities in a real academic environment. In the first phase of the experiment, the participants reported daily changes in their sleep measures and self-evaluated emotional well-being throughout a one-semester academic term, using the Daily Sampling System smartphone application. The psychometric data analysis demonstrates a significant longitudinal correlation of the subjective level of fatigue with stress, daytime sleepiness, sleep duration, and alertness on awakening, which is additionally characterized by a strong inter-subject variability. In the second phase, the participants were invited biweekly for resting-state electroencephalography and electrocardiography recordings that complemented filling out the Depression, Anxiety, and Stress Scales. Spectral analysis of the electroencephalographic samples discovers significant alterations that accompany emotional state changes, in particular, an intensification in the low-frequency neural oscillatory activity while being in a state of anxiety or stress. Following the resting-state data recording, the subjects of this study took part in a specially designed task for sustained attention during their regular classroom activities, to investigate associations of the non-event-related brain activity spectral properties with the level of alertness to randomly introduced visual stimuli. Finally, the data collected in this experiment were used for training a machine learning classification system, indicating the possibility of developing an implicit brain-computer interface aimed for recognizing the states of inattention and emotional arousal of its users after subject-specific calibration.