臺灣博碩士論文加值系統

隨著老年人口快速成長，相對失能人口也將大幅增加，其所導致的長期照顧需求也隨之遽增。因此在長期照顧體系中，專業照顧服務員與家庭照顧者（兩者以下簡稱照顧者）為最重要且最直接的照顧人力。
照顧者的工作內容重覆性高，且有些照顧動作需要負重或長時間過度施力，其中以移動換位最容易造成肌肉骨骼損傷，而目前用來量測肌肉骨骼損傷的工具為肌電圖（EMG），它是一種利用針極刺入肌肉，記錄其各種狀態下的電位活動，協助醫生診斷肌肉或神經的異常活動，肌電圖檢查的優點為準確度高，但時間及人力成本高，且該項檢查為侵入性檢查，故有一定程度之風險。
因此本研究提出透過 AI 神經網路關鍵點辨識技術做影像分析，並以站姿轉移位的動作姿勢、病床高度及被照顧者體重這三個變因做探討，進一步了解在已知的三個變因之下，透過 AI 影像分析是否可以辨別這三個變因將造成照顧者肌肉骨骼損傷。依據本研究實驗結果可以得知 AI 能辨別且相較於EMG在使用上更為方便且時間及人力成本低，最重要的是無須侵入性檢查， AI 還可即時記錄本文所提出的人體十七個關鍵點同時施力時，關鍵點的細微變化及每日轉移位的動作次數。

With population aging, chronic diseases and organ malfunctioning are wide spreading, which correspond to increasing disabled people. The long-term care system needed requires professional caregiver and family caregiver (both termed “caregiver” hereafter) as its major manpower.
In the repetitive labor from the caregivers, some require great weight loading or prolonged excessive force applying, among which mobile transposition most readily causes musculoskeletal injuries. Electromyography (EMG) is a present technique to measure these injuries, through recording the various electrode potentials that help a physician diagnose abnormal muscular or neural activities using muscle-piercing electrodes. While being relatively accurate, the EMG has the disadvantages of being time- labor-consuming and invasive.
The author proposes a methodology utilizing the image analyses through the artificial intelligence (AI) that recognizes the critical points in neural networks. Taking the moving gesture, bed height, and the body weight of the care recipient as key variables, the AI image analyses were implemented to distinguish which variable leads the musculoskeletal injuries to the caregivers.
The experimental results showed that the AI is capable of discerning the key variable in a more convenient, time-labor-efficient, and non-invasive way. The AI realtime records the subtle changes and the number of times of daily transpositions during the force-applying from the 17 critical points on the caregivers.

摘 要 i
ABSTRACT ii
目 錄 iv
圖目錄 vi
表目錄 viii
第一章 導論 1
1.1研究背景 1
1.2研究動機 4
1.3研究目的 5
第二章 文獻探討 6
2.1 移動換位造成的傷害因子 6
2.1.1 工作姿勢-動作 6
2.1.2 工作姿勢-高度 7
2.1.2 工作姿勢-重量 7
2.2 肌肉骨骼損傷 8
2.3 家庭照顧者現況 10
2.4 深度學習神經網路 12
2.4.1 卷積神經網絡(CNN) 13
2.4.2 區域卷積神經網絡(R-CNN) 14
2.4.3 快速區域卷積神經網絡(Fast R-CNN) 14
2.4.4 更快速區域卷積神經網絡(Faster R-CNN) 15
2.4.5 Mask R-CNN 16
2.4.6 Keypoint RCNN 17
2.5 關鍵點辨識 17
第三章 研究方法 19
3.1關鍵點辨識應用架構 19
3.2影像辨識實際導入的難處及限制 22
3.3 實驗設計及流程（DOE） 22
3.4影像數據校正 25
3.5影像數據離群值剔除 26
3.6影像數據分段擷取 26
第四章 研究結果與討論 27
4.1 站姿轉移位 <實驗一與實驗五> 27
4.2 站姿轉移位 <實驗二與實驗六> 31
4.3 站姿轉移位 <實驗三與實驗七> 34
4.4 站姿轉移位 <實驗四與實驗八> 36
4.5 統計檢定 <變異數分析Analysis of variance, ANOVA> 39
4.6 討論 41
第五章 結論與未來研究 42
第六章 參考文獻 43

[1] Liu, W., Wang, Z., Liu, X., Zeng, N., Liu, Y., & Alsaadi, F. E. (2017). A survey of deep neural network architectures and their applications. Neurocomputing, 234, 11-26.
[2] Yu, D., & Deng, L. (2010). Deep learning and its applications to signal and information processing [exploratory dsp]. IEEE Signal Processing Magazine, 28(1), 145-154.
[3] Graves, A., Mohamed, A. R., & Hinton, G. (2013, May). Speech recognition with deep recurrent neural networks. In 2013 IEEE international conference on acoustics, speech and signal processing (pp. 6645-6649). Ieee.
[4] Young, T., Hazarika, D., Poria, S., & Cambria, E. (2018). Recent trends in deep learning based natural language processing. ieee Computational intelligenCe magazine, 13(3), 55-75.
[5] Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 25, 1097-1105.
[6] Nielsen, M. A. (2015). Neural networks and deep learning (Vol. 25). San Francisco, CA: Determination press.
[7] Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., Van Den Driessche, G., ... & Hassabis, D. (2016). Mastering the game of Go with deep neural networks and tree search. nature, 529(7587), 484-489.
[7] Darragh, A. R., Sommerich, C. M., Lavender, S. A., Tanner, K. J., Vogel, K., & Campo, M. (2015). Musculoskeletal discomfort, physical demand, and caregiving activities in informal caregivers. Journal of Applied Gerontology, 34(6), 734-760.
[8] Vincent-Onabajo, G., Daniel, H., Aliyu Lawan, M. U. A., Masta, M. A., & Modu, A. (2018). Musculoskeletal symptoms among family caregivers of community-dwelling stroke survivors in Nigeria. Journal of caring sciences, 7(2), 59.
[9] Johnsson, C., Kjellberg, K., Kjellberg, A., Lagerström, M.(2004). A direct observation instrument for assessment of nurses' patient transfer technique (DINO). Applied Ergonomics, 35(6), 591-601.
[10] Adler, R., & Mehta, R. (2014). Catalyzing technology to support family caregiving. National Alliance for Caregiving, 1-18.
[11] C. Vijayalakshmi and N. Kumar, "Survey on Risk Estimation of Chronic Disease using Machine Learning," (in English), 2019.

[12] Ibrahim, Y., Wang, H., Liu, J., Wei, J., Chen, L., Rech, P., ... & Guo, G. (2020). Soft errors in DNN accelerators: A comprehensive review. Microelectronics Reliability, 115, 113969.
[13] Gibney, E. (2016). Google AI algorithm masters ancient game of Go. Nature News, 529(7587), 445.
[14] Bengio, Y. (2009). Learning deep architectures for AI. Now Publishers Inc.
[15] Fukushima, K., & Miyake, S. (1982). Neocognitron: A self-organizing neural network model for a mechanism of visual pattern recognition. In Competition and cooperation in neural nets (pp. 267-285). Springer, Berlin, Heidelberg.
[16] LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.
[17] Jones, N. (2014). Computer science: The learning machines. Nature News, 505(7482), 146.
[18] He, K., Gkioxari, G., Dollár, P., & Girshick, R. (2017). Mask r-cnn. In Proceedings of the IEEE international conference on computer vision (pp. 2961-2969).
[19] Toshev, A., & Szegedy, C. (2014). Deeppose: Human pose estimation via deep neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1653-1660).
[20] Engels, J. A., Van Der Gulden, J. W. J., Senden, T. F., Hertog, C. A. W. M., Kolk, J. J., & Binkhorst, R. A. (1994). Physical work load and its assessment among the nursing staff in nursing homes. Journal of Occupational Medicine, 338-345.
[21] Darragh, A. R., Sommerich, C. M., Lavender, S. A., Tanner, K. J., Vogel, K., & Campo, M. (2015). Musculoskeletal discomfort, physical demand, and caregiving activities in informal caregivers. Journal of Applied Gerontology, 34(6), 734-760.
[22] Adler, R., & Mehta, R. (2014). Catalyzing technology to support family caregiving. National Alliance for Caregiving, 1-18.（Adler, R., & Mehta, R.，2014）
[23] Wright Jr, K. P., Bogan, R. K., & Wyatt, J. K. (2013). Shift work and the assessment and management of shift work disorder (SWD). Sleep medicine reviews, 17(1), 41-54.
[24] HSA. (2011). Guidance on the Management of Manual Handling in Healthcare.
[25] Suykens, J. A., & Vandewalle, J. (1999). Least squares support vector machine classifiers. Neural processing letters, 9(3), 293-300.
[26] Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 580-587).
[27] Girshick, R. (2015). Fast r‐cnn In Proceedings of the IEEE International Conference on Computer Vision (pp. 1440–1448). Piscataway, NJ: IEEE.[Google Scholar].


[28] Ren, S., He, K., Girshick, R., & Sun, J. (2015). Faster r-cnn: Towards real-time object detection with region proposal networks. Advances in neural information processing systems, 28, 91-99.
[29] Zhou, X., Gong, W., Fu, W., & Du, F. (2017, May). Application of deep learning in object detection. In 2017 IEEE/ACIS 16th International Conference on Computer and Information Science (ICIS) (pp. 631-634). IEEE.
[30] Long, J., Shelhamer, E., & Darrell, T. (2015). Fully convolutional networks for semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3431-3440).
[31] Pinho, R. R., & Tavares, J. M. R. (2009). Tracking features in image sequences with kalman filtering, global optimization, mahalanobis distance and a management model.
[32] Pinho, R. R., Tavares, J. M. R., & Correia, M. V. (2006). An improved management model for tracking missing features in computer vision long image sequences.
[33] Pinho, R. R., Tavares, J. M. R., & Correia, M. V. (2005). A movement tracking management model with Kalman filtering, global optimization techniques and mahalanobis distance. Advances in Computational Methods in Sciences and Engineering 2005, Vols 4 A & 4 B
[34] Tavares, J. M. R. S., & Padilha, A. J. M. N. (1995). Matching Lines in Image Sequences using Geometric Constraints. Proceedings RecPad'95, Aveiro.
[35] Zhu, Y., Dariush, B., & Fujimura, K. (2008, June). Controlled human pose estimation from depth image streams. In 2008 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (pp. 1-8). IEEE.
[36] Sigalas, M., Pateraki, M., & Trahanias, P. (2015). Full-body pose tracking—the top view reprojection approach. IEEE transactions on pattern analysis and machine intelligence, 38(8), 1569-1582.
[37] Zuffi, S., & Black, M. J. (2015). The stitched puppet: A graphical model of 3d human shape and pose. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 3537-3546).
[38] Ye, M., Wang, X., Yang, R., Ren, L., & Pollefeys, M. (2011, November). Accurate 3d pose estimation from a single depth image. In 2011 International Conference on Computer Vision (pp. 731-738). IEEE.
[39] Shotton, J., Girshick, R., Fitzgibbon, A., Sharp, T., Cook, M., Finocchio, M., ... & Blake, A. (2012). Efficient human pose estimation from single depth images. IEEE transactions on pattern analysis and machine intelligence, 35(12), 2821-2840.
[40] Woldegiorgis, B. H., Lin, C. J., & Sananta, R. (2021). Using Kinect body joint detection system to predict energy expenditures during physical activities. Applied Ergonomics, 97, 103540.（Woldegiorgis, B. H., Lin, C. J., & Sananta, R.，2021）
[41] Andriluka, M., Pishchulin, L., Gehler, P., Schiele, B., 2018b. MPII human pose database. URL http://human-pose.mpi-inf.mpg.de/#results.
[42] Wei, S. E., Ramakrishna, V., Kanade, T., & Sheikh, Y. (2016). Convolutional pose machines. In Proceedings of the IEEE conference on Computer Vision and Pattern Recognition (pp. 4724-4732).（
[43] Cao, Z., Simon, T., Wei, S. E., & Sheikh, Y. (2017). Realtime multi-person 2d pose estimation using part affinity fields. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7291-7299).
[44] Newell, A., Huang, Z., & Deng, J. (2016). Associative embedding: End-to-end learning for joint detection and grouping. arXiv preprint arXiv:1611.05424.
[45] Impact of mental health and caregiver burden on family caregivers’ physical health. Archives of gerontology and geriatrics, 50(3), 267-271
[46] Collins, J. W., Wolf, L., Bell, J., & Evanoff, B. (2004). An evaluation of a “best practices” musculoskeletal injury prevention program in nursing homes. Injury Prevention, 10(4), 206-211.（Collins, J. W., Wolf, L., Bell, J., & Evanoff, B.，2004）
[47] Hignett, S., Fray, M., Rossi, M. A., Tamminen-Peter, L., Hermann, S., Lomi, C., ... & Johnsson, C. (2007). Implementation of the Manual Handling Directive in the healthcare industry in the European Union for patient handling tasks. International Journal of Industrial Ergonomics, 37(5), 415-423.
[48] Ngan, K., Drebit, S., Siow, S., Yu, S., Keen, D., & Alamgir, H. (2010). Risks and causes of musculoskeletal injuries among health care workers. Occupational medicine, 60(5), 389-394.
[49] Park, R. M., Bushnell, P. T., Bailer, A. J., Collins, J. W., & Stayner, L. T. (2009). Impact of publicly sponsored interventions on musculoskeletal injury claims in nursing homes. American journal of industrial medicine, 52(9), 683-697.
[50] Executive, H. a. S. (2009-10). Health and Safety Executive. Pesticides—The Blue Book, available electronically at http://www. pesticides. gov.uk/Blue_Book/Contents.
[51] Ontario.ca. (2011). Prevent Musculoskeletal Disorders (MSDs) in Health Care Workplaces. Retrieved from http://www.labour.gov.on.ca/english/hs/sawo/pubs/fs_msd_healthcare.php.
[52] Statistics, B. o. L. (2012). Incidence rates and numbers of nonfatal occupational injuries by selected industries and ownership.




[53] Commission, E. (2010). Communication from the Commission to the European Parliament Pursuant to Article 294 (6) of the Treaty on the Functioning of the European Union Concerning the Position Adopted by the Council with a View to the Adoption of a Proposal for a Regulation of the European Parliament and of the Council Establishing a European Asylum Support Office and a Proposal for a Decision of the European Parliament and of the Council Amending Decision 573/2007/EC of the European Parliament and of the Council of 23 May 2007 Establishing the European Refugee Fund for the Period 2008 to 2013 by Removing Funding for Certain Community Actions and Altering the Limit for Funding Such Actions: Publications Office.
[54] Suka, M., & Yoshida, K. (2005). Musculoskeletal pain in Japan: prevalence and interference with daily activities. Modern Rheumatology, 15(1), 41-47.
[55] 內政部統計處(2021)。內政部人口統計資料01 縣市人口按性別及五齡組(63)、2021年9月1日，取自httt:://sowf.moi.gov.tw/stat/.ean/list.ttm
[56] 國家發展委員會(2020)。中華民國人口推估（2020至2070年），2021年9月23日，取自https://www.ndc.gov.tw/Content_List.aspx?n=695E69E28C6AC7F3。
[57] 行政院衛生署(2010)。國民長期照護需要調查。臺北：行政院衛生署。
[58] 行政院經濟建設委員會(2006)，中華民國臺灣95 年至140 年人口推計，臺北市：行政院經濟建設委員會。
[59] 國家發展委員會(2012)‧中華民國2012年至2060年人口推計(中推計)‧取自http://goo.gl/LJruw6
[60] 曲同光, 崔道華, 彭美琪, & 陳信婷. (2015). 我國長期照顧保險制度規劃概述. 長庚科技學刊, (23), 1-14.
[61] 胡幼慧、王孝先、郭淑珍(1995)。〈家人照護失能老人的困境：一項質化與量化整合的研究〉，《公共衛生》，22(2)：99-113（胡幼慧、王孝先、郭淑珍，1995）
[62] 邱啟潤(2002)。高雄市居家照護個案的主要照顧者綜合性需求。長青研究發展通訊，16，1-5
[63] 張淑卿, & 陸子初. (2019). 國際失能者家庭照顧現況與支持策略. 長期照護雜誌, 23(1), 1-10.
[64] 邱麗蓉、謝佳容、蔡欣玲(2007)。失智症病患主要照護者的壓力源、評價和因應行為與健康之相關性探討。精神衛生護理雜誌，2(2)，31-44
[65] 周月清。（2002）。台灣障礙者及老人社區照顧—國家、地方政府、與家庭照護之角色為何？東吳大學社會工作學報，8，19-74。（外審學術期刊）（NSC-88-2412-H-031-003）
[66] 郭外天, & 張瀞仁. (2014). 安全照護之國際趨勢:[No-Lift Policy] 不徒手搬運病患規範. 長期照護雜誌, 18(1), 29-39.
[67] 李禹璇, 翁瑞萱, 徐雅媛, & 王子娟. (2013). 勞工工作環境與自覺肌肉骨骼不適之相關研究. 勞工安全衛生研究季刊, 21(4), 432-441.
[68] 楊忠一. (2014). 轉移位輔具使用概念與國內應用概況. 長期照護雜誌, 18(1), 49-58.
[69] 陳明山、陳志勇。(2007)。看護人員肌肉骨骼傷害調查與作業方法改善。台北市：勞動部勞動及職業安全衛生研究所。
[70] 陳雅芬. (2022). 應用AI影像辨識技術蒐集人體關鍵於高齡平衡功能評估之研究. (碩士). 元智大學, 桃園縣.
[71] 鄭誠功. (1999). 生物力學評估與應用研究—下背傷害力學模式於現場實證改善研究. 行政院勞工委員會勞工安全衛生研究所
[72] 潘儀聰. (2011). 人因工程改善一舉數得. 勞工安全衛生簡訊. 105:9-11
[73] 潘儀聰、游志雲. (2008)，人因工程現場不良工作姿勢改善績效評估研究，行政院勞工委員會勞工安全衛生研究所.