臺灣博碩士論文加值系統

背景：
心理壓力是現代人在生活步調快速、變化多端的環境中所面對的問題，長期處於壓力環境會對生心理造成影響。過去已證實心電訊號 (Electrocardiography, ECG) 計算之心率變異性 (Heart rate variability, HRV) 及肌電訊號 (Electromyography, EMG) 共同作為壓力指標可以有效提升壓力的辨識程度。然而過去量測多種訊號會使用不同儀器，且穿戴裝置之雜訊品質指標仍無一致性的閾值。

假說：
使用單一穿戴設備紀錄ECG，開發透過雜訊品質指標去除雜訊，並將產生的EMG分離且保留，此方式能減少儀器使用，提升壓力偵測的準確度。

材料與方法：
以實驗室開發之貼片型心率變異分析儀黏貼於受試者右上斜方肌，量測5分鐘的基礎值後，隨機給予三種壓力實驗，包含5分鐘史楚普實驗、5分鐘算數實驗以及2分鐘演講實驗，於每項實驗結束後填寫壓力視覺量表 (Visual analogue scale of stress, VAS of stress)。本研究共開發三種演算法，包含雜訊品質指標計算、R波偵測演算法及ECG與EMG分離演算法。透過計算HRV及EMG之時域及頻域參數，使用隨機森林 (Random forest) 分類壓力情緒並預測主觀壓力程度。

結果：
雜訊品質指標能透過閾值有效排除雜訊，R波偵測演算法使用於MIT-BIH心電資料庫準確度達97%、敏感度99%，ECG與EMG分離演算法能保留兩種訊號的特徵，反映出交感神經活性及肌肉收縮變化。ECG心跳間隔 (R-R interval, RRI) 之平均值在壓力情境中皆顯著下降，RRI之峰值顯著上升，與交感神經活性有關之LF%、LF/HF以及與肌肉變化有關之平均平方數及平均絕對值皆顯著上升。特徵挑選後發現，同時使用ECG及EMG特徵比單獨使用一種訊號特徵所建立之模型，有更好的辨識結果，準確度達84%。

結論：
本研究透過單一穿戴裝置，開發以使用最少儀器的方式，建立準確度高之心理壓力演算法。

Background:
Psychological stress is a common issue that effects people in this fast-paced society and rapidly changing working environment. A current study showed that using the combination of heart rate variability (HRV), which is measured by electrocardiography (ECG), and electromyography (EMG), the ability of stress detection increases significantly. However, the collection of ECG and EMG often come from different devices, which acquire numerous electrodes used on each individual. Moreover, the indices of signal quality from ECG-detecting wearable device have no standard threshold.

Hypotheses:
The proposed algorithm based on quality indices for reducing the noise signal can divide EMG from a single wearable ECG device. It reduced the usage of electrodes and had higher accuracy than multi-device.

Materials and Methods:
The subjects underwent three stress-increasing tasks: five minutes of Stroop color-word test, five minutes of the arithmetic test, and two minutes of a random speech test; and a visual analogue scale of stress (VAS of stress) was asked to be filled after three tasks ended. The physiological signals were collected throughout the tasks by a laboratory-made wearable device and from the subjects’ trapezius. Three algorithms were developed in this study: signal quality indices calculation, R peak detection, and separation of EMG from ECG. This study calculated time domain and frequency domain features from HRV and EMG for classifying stressful emotions by a random forest model.

Results:
According to this study, signal quality indices by a threshold can exclude noise. The R peak detection algorithm achieved an average accuracy of 97% and sensitivity of 99% for the MIT-BIH ECG database. The proposed algorithm for the separation of EMG from ECG reserved the features of ECG and EMG, which demonstrated the activities of the autonomic nervous system and muscles. The mean of the R-R interval (RRI) significantly decreased during stressful situations, while the kurtosis of RRI significantly increased. The normalized low frequency and the low frequency to high frequency power ratio, which represents the sympathetic nervous system, significantly increased. Root mean square of EMG and mean absolute value of EMG, which represents the trapezius activity, also significantly increased. The combination of HRV and EMG features had the highest performance among all features. The accuracy of stress recognition in two levels was 84%.

Conclusions:
The proposed algorithm can reduce the number of sensors and has high accuracy for psychological stress detection.

致謝 i
中文摘要 ii
英文摘要 iii
目錄 v
表目錄 vii
圖目錄 viii
第壹章、緒論 1
第一節、壓力介紹 1
第二節、心電訊號與肌電訊號介紹 2
第三節、壓力指標與生理反應 2
第四節、生理訊號處理演算法 4
第五節、壓力偵測之研究及應用 5
第貳章、研究假說與目的 6
第一節、問題與假說 6
第二節、研究目的 6
第參章、研究材料與方法 7
第一節、研究對象 7
第二節、實驗流程 7
第三節、研究工具 8
第四節、演算法建立 9
第五節、特徵擷取 12
第六節、統計方法 14
第七節、模型訓練 14
第八節、驗證方法 15
第肆章、研究結果 19
第一節、受試者基本資料及問卷統計結果 19
第二節、演算法開發結果 19
第三節、生理參數統計結果 20
第四節、模型訓練結果 21
第伍章、討論與結論 23
第一節、本研究之重要發現 23
第二節、演算法開發討論 23
第三節、壓力對心電訊號及肌電訊號的影響 24
第四節、主觀壓力感受的相關性 24
第五節、模型訓練成效 25
第六節、研究限制 25
第七節、研究貢獻 26
第八節、研究未來展望 26
第九節、結論 27
參考文獻 28
附表 33
附圖 46
附錄 77
附錄一、人體研究計畫申請審查證明 77

表目錄
表一、貼片型心率變異分析儀之規格 33
表二、心率變異性參數 34
表三、肌電訊號參數 35
表四、受試者基本資料 36
表五、R波偵測演算法驗證 37
表六、比較不同特徵選取之二元分類模型評估結果 40
表七、比較不同特徵選取之多元分類模型評估結果 41
表八、比較不同特徵選取之多迴歸模型評估結果 43
表九、比較壓力偵測模型之研究 45

圖目錄
圖一、儀器黏貼位置 46
圖二、實驗流程圖 47
圖三、貼片型心率變異分析儀之資訊 48
圖四、訊號分析演算法流程圖 50
圖五、頻域分析流程圖 51
圖六、機器學習模型訓練流程圖 52
圖七、壓力視覺量表分數在基礎值及三種壓力實驗下之差異 53
圖八、計算訊號品質指標之結果 54
圖九、R波偵測演算法過程 55
圖十、心電訊號及肌電訊號之分離演算法結果 56
圖十一、心率變異性參數在不同壓力情境下之差異 58
圖十二、肌電參數在不同壓力情境下之差異 60
圖十三、心率變異性參數與壓力視覺量表分數之相關性 62
圖十四、肌電參數與壓力視覺量表分數之相關性 64
圖十五、使用混淆矩陣評估特徵挑選後之二元分類模型結果 65
圖十六、使用ROC curve評估特徵挑選後之二元分類模型辨識率 66
圖十七、使用混淆矩陣評估特徵挑選後之多元分類模型結果 68
圖十八、使用ROC curve評估特徵挑選後之多元分類模型辨識率 70
圖十九、使用殘差圖及直方圖呈現訓練集特徵挑選後之迴歸模型結果 72
圖二十、使用殘差圖及直方圖呈現測試集特徵挑選後之迴歸模型結果 74
圖二十一、使用Bland-Altman plot評估迴歸模型實際值與預測值之一致性 76

1. Won, E. and Y.K. Kim, Stress, the Autonomic Nervous System, and the Immune-kynurenine Pathway in the Etiology of Depression. Curr Neuropharmacol, 2016. 14(7): p. 665-73.
2. Goldstein, D.S. and I.J. Kopin, Evolution of concepts of stress. Stress, 2007. 10(2): p. 109-20.
3. Rosmond, R. and P. Bjorntorp, Endocrine and metabolic aberrations in men with abdominal obesity in relation to anxio-depressive infirmity. Metabolism, 1998. 47(10): p. 1187-93.
4. McMartin, C., et al., Receipt of 5-Alpha Reductase Inhibitors Before Radical Cystectomy: Do They Render High-Grade Bladder Tumors Less Aggressive? Clin Genitourin Cancer, 2019. 17(6): p. e1122-e1128.
5. McEwen, B.S., Protection and damage from acute and chronic stress: allostasis and allostatic overload and relevance to the pathophysiology of psychiatric disorders. Ann N Y Acad Sci, 2004. 1032: p. 1-7.
6. McEwen, B.S., The neurobiology of stress: from serendipity to clinical relevance. Brain Res, 2000. 886(1-2): p. 172-189.
7. Brown, M.R., et al., Corticotropin-releasing factor: a physiologic regulator of adrenal epinephrine secretion. Brain Res, 1985. 328(2): p. 355-7.
8. Brotman, D.J., S.H. Golden, and I.S. Wittstein, The cardiovascular toll of stress. Lancet, 2007. 370(9592): p. 1089-100.
9. Heart rate variability. Standards of measurement, physiological interpretation, and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Eur Heart J, 1996. 17(3): p. 354-81.
10. Castaldo, R., et al., Acute mental stress assessment via short term HRV analysis in healthy adults: A systematic review with meta-analysis. Biomedical Signal Processing and Control, 2015. 18: p. 370-377.
11. Salahuddin, L., et al., Ultra short term analysis of heart rate variability for monitoring mental stress in mobile settings. Annu Int Conf IEEE Eng Med Biol Soc, 2007. 2007: p. 4656-9.
12. Kirschbaum, C., K.M. Pirke, and D.H. Hellhammer, The 'Trier Social Stress Test'--a tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology, 1993. 28(1-2): p. 76-81.
13. Coutts, L.V., et al., Deep learning with wearable based heart rate variability for prediction of mental and general health. Journal of Biomedical Informatics, 2020. 112.
14. Wozniak, H., et al., Mental health outcomes of ICU and non-ICU healthcare workers during the COVID-19 outbreak: a cross-sectional study. Annals of Intensive Care, 2021. 11(1): p. 106.
15. Pan, J. and W.J. Tompkins, A real-time QRS detection algorithm. IEEE Trans Biomed Eng, 1985. 32(3): p. 230-6.
16. Manikandan, M.S. and K.P. Soman, A novel method for detecting R-peaks in electrocardiogram (ECG) signal. Biomedical Signal Processing and Control, 2012. 7(2): p. 118-128.
17. Chen, R., et al., Needle EMG of the human diaphragm: power spectral analysis in normal subjects. Muscle Nerve, 1996. 19(3): p. 324-30.
18. Mello, R.G., L.F. Oliveira, and J. Nadal, Digital Butterworth filter for subtracting noise from low magnitude surface electromyogram. Comput Methods Programs Biomed, 2007. 87(1): p. 28-35.
19. Redfern, M.S., R.E. Hughes, and D.B. Chaffin, High-Pass Filtering to Remove Electrocardiographic Interference from Torso Emg Recordings. Clinical Biomechanics, 1993. 8(1): p. 44-48.
20. Ren, X.M., et al., Noise reduction based on ICA decomposition and wavelet transform for the extraction of motor unit action potentials. Journal of Neuroscience Methods, 2006. 158(2): p. 313-322.
21. Hellhammer, D.H., S. Wust, and B.M. Kudielka, Salivary cortisol as a biomarker in stress research. Psychoneuroendocrinology, 2009. 34(2): p. 163-171.
22. Heart Rate Variability. Annals of Internal Medicine, 1993. 118(6): p. 436-447.
23. Pourmohammadi, S. and A. Maleki, Stress detection using ECG and EMG signals: A comprehensive study. Comput Methods Programs Biomed, 2020. 193: p. 105482.
24. Wijsman, J., et al., Trapezius Muscle EMG as Predictor of Mental Stress. Acm Transactions on Embedded Computing Systems, 2013. 12(4).
25. Phinyomark, A., P. Phukpattaranont, and C. Limsakul, Feature reduction and selection for EMG signal classification. Expert Systems with Applications, 2012. 39(8): p. 7420-7431.
26. Li, Q., R.G. Mark, and G.D. Clifford, Robust heart rate estimation from multiple asynchronous noisy sources using signal quality indices and a Kalman filter. Physiol Meas, 2008. 29(1): p. 15-32.
27. Redmond, S.J., et al., ECG quality measures in telecare monitoring. Annu Int Conf IEEE Eng Med Biol Soc, 2008. 2008: p. 2869-72.
28. Rahman, S., et al., Robustness of electrocardiogram signal quality indices. J R Soc Interface, 2022. 19(189): p. 20220012.
29. Kim, J., J. Park, and J. Park, Development of a statistical model to classify driving stress levels using galvanic skin responses. Human Factors and Ergonomics in Manufacturing & Service Industries, 2020. 30(5): p. 321-328.
30. Pakhomov, S.V.S., et al., Using consumer-wearable technology for remote assessment of physiological response to stress in the naturalistic environment. PLoS One, 2020. 15(3): p. e0229942.
31. Saganowski, S., et al., Review of consumer wearables in emotion, stress, meditation, sleep, and activity detection and analysis. arXiv preprint arXiv:2005.00093, 2020.
32. Ahn, J.W., Y. Ku, and H.C. Kim, A novel wearable EEG and ECG recording system for stress assessment. Sensors, 2019. 19(9): p. 1991.
33. Can, Y.S., B. Arnrich, and C. Ersoy, Stress detection in daily life scenarios using smart phones and wearable sensors: A survey. J Biomed Inform, 2019. 92: p. 103139.
34. Posada-Quintero, H.F. and J.B. Bolkhovsky, Machine Learning models for the Identification of Cognitive Tasks using Autonomic Reactions from Heart Rate Variability and Electrodermal Activity. Behavioral Sciences, 2019. 9(4).
35. Moody, G.B. and R.G. Mark, The impact of the MIT-BIH arrhythmia database. IEEE Eng Med Biol Mag, 2001. 20(3): p. 45-50.
36. Goldberger, A.L., et al., PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation, 2000. 101(23): p. E215-20.
37. Castaldo, R., et al. To What Extent Can We Shorten HRV Analysis in Wearable Sensing? A Case Study on Mental Stress Detection. 2018. Singapore: Springer Singapore.
38. Smets, E., et al. Comparison of Machine Learning Techniques for Psychophysiological Stress Detection. 2016. Cham: Springer International Publishing.
39. Choi, J., B. Ahmed, and R. Gutierrez-Osuna, Development and Evaluation of an Ambulatory Stress Monitor Based on Wearable Sensors. IEEE Transactions on Information Technology in Biomedicine, 2012. 16(2): p. 279-286.
40. Clifford, G.D., et al., Signal quality indices and data fusion for determining clinical acceptability of electrocardiograms. Physiol Meas, 2012. 33(9): p. 1419-33.
41. Kužílek, J., et al. Data driven approach to ECG signal quality assessment using multistep SVM classification. in 2011 Computing in Cardiology. 2011.
42. Zhang, Q., L. Fu, and L. Gu, A Cascaded Convolutional Neural Network for Assessing Signal Quality of Dynamic ECG. Comput Math Methods Med, 2019. 2019: p. 7095137.
43. Visnovcova, Z., et al., Complexity and time asymmetry of heart rate variability are altered in acute mental stress. Physiol Meas, 2014. 35(7): p. 1319-34.
44. Schubert, C., et al., Effects of stress on heart rate complexity--a comparison between short-term and chronic stress. Biol Psychol, 2009. 80(3): p. 325-32.
45. Hagberg, C. and M. Hagberg, Surface EMG amplitude and frequency dependence on exerted force for the upper trapezius muscle: a comparison between right and left sides. Eur J Appl Physiol Occup Physiol, 1989. 58(6): p. 641-5.
46. Choi, J., B. Ahmed, and R. Gutierrez-Osuna, Development and evaluation of an ambulatory stress monitor based on wearable sensors. IEEE Trans Inf Technol Biomed, 2012. 16(2): p. 279-86.