簡易檢索 / 詳目顯示

研究生: 陳姿樺
Chen, Tzu-Hua
論文名稱: 生理訊號及聲學特徵作為探討大專院校生心理健康與情緒狀態指標之前驅研究:以橫斷研究為例
A Preliminary Cross-Sectional Study on Physiological Signals and Acoustic Features as Indicators of Mental Health and Emotional States in University Students
指導教授: 李昆樺
Lee, Kun-Hua
口試委員: 劉奕汶
LIU, YI-WEN
張芸瑄
學位類別: 碩士
Master
系所名稱: 竹師教育學院 - 教育心理與諮商學系
Educational Psychology and Counseling
論文出版年: 2025
畢業學年度: 113
語文別: 中文
論文頁數: 51
中文關鍵詞: 聲學特徵心率變異性腦波大專院校生心理健康
外文關鍵詞: Acoustic Features, Heart Rate Variability, Electroencephalography, College Student, Mental Health
相關次數: 點閱:31下載:2
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  • 本研究旨在探討大專院校學生的心理健康問題,並探索使用生理訊號如腦波(EEG)、心率變異性(HRV),與各類聲學特徵來評估情緒狀態的可行性。
    研究目的是為瞭解生理訊號和聲學特徵在正常情緒與憂鬱症狀上的差異。而本研究
    為一項初步探索,盼為臨床診斷提供新的可能性,進而促進心理問題之早期預防與介入。
    研究方法結合問卷調查、生理訊號測量以及聲學實驗,採用階段實驗設計,收集受
    試者在基準狀態和情緒誘導後的聲學特徵。研究透過網路招募17位年齡介於20至30
    歲的大專院校生。研究中使用工具有基本資料問卷、貝克憂鬱量表第二版(BDI-II),並得憂鬱量表總平均分數10.12。資料整合上運用Praat和Librosa軟體來計算聲學特徵數據,統計方法則採皮爾森積差相關和線性迴歸,以分析生理訊號、聲學特徵與量表分數之間的相關性。
    研究結果顯示,腦波(Beta、Delta、Gamma)及心率變異性(LF、HF、LF/HF)在不同階段與憂鬱症狀無顯著相關。然而,在情緒誘導階段中,聲學特徵中的第三共振峰(F3)最小值與貝克憂鬱量表(BDI-II)得分呈顯著負相關(r = -.58, p < .05),並能顯著預測憂鬱症狀(R² = .336, p = .015)。本研究初步發現,F3 最小值為唯一具顯著相關且具預測力的變項,顯示其作為憂鬱症狀指標之潛力。
    研究建議,未來可以擴大樣本範圍與數量,並納入不同文化背景的學生群體。研究
    應發展更標準化且客觀的情緒誘導方法,以確保實驗程序的一致性與結果可信度。此外,可擴展研究範疇至更多元的生理訊號與聲學特徵指標,並結合機器學習等進階分析技術,以建構更具預測效度的情緒評估模型。研究也建議結合生理訊號與聲學特徵指標,以提升對情緒狀態的客觀測量。研究結果可應用於建立大專院校新生入學時的心理健康篩檢機制,也鼓勵臨床工作者將生理與聲學特徵納入評估工具,輔助傳統心理測量與診斷。


    This study explored the use of physiological signals (EEG, HRV) and acoustic features to assess emotional states and identify depressive symptoms in university students. This preliminary research aims to enhance early diagnosis and intervention for mental health issues.
    Seventeen university students (aged 20-30) participated in a phased experiment involving questionnaires (BDI-II), physiological measurements, and acoustic recordings at baseline and during emotion induction. Data was analyzed using Praat and Librosa for acoustic features, and Pearson correlation and linear regression for statistical analysis.
    Results showed no significant correlation between EEG (Beta, Delta, Gamma) or HRV (LF, HF, LF/HF) and depressive symptoms. However, during emotion induction, the minimum value of the third formant (F3_min), an acoustic feature, significantly negatively correlated with BDI-II scores (r=−.58,p<.05) and significantly predicted depressive symptoms (R2=.336,p=.015). This suggests F3min's potential as a depressive symptom indicator.
    Future research should expand sample size and diversity, standardize emotion-induction methods, and integrate more physiological and acoustic indicators with machine learning for improved predictive models. Combining physiological and acoustic features could enhance objective emotional assessment. Findings could be applied to mental health screening for university students and encourage clinicians to incorporate these tools into their assessment practices.

    QR CODE