睡眠姿勢(以下簡稱睡姿)是評估睡眠品質的重要因素之一，同時也與許多慢性疾病有所關聯。在醫療實務中，睡姿對於睡眠呼吸中止症影響重大，其患者應避免仰臥的睡姿，並儘量以側身姿勢入眠。除了睡眠呼吸中止症之外，睡姿亦與壓瘡的防治有關。所謂壓瘡是指人體覆蓋骨突處的皮膚長期受到外在壓力，造成皮膚、皮下組織、肌肉與骨頭受傷、潰瘍甚至壞死。剛完成手術及行動不便需長期臥床的病患因無法自行翻身，而容易造成壓瘡產生，這些病患需要照護人員定期幫他們翻身，以避免長期維持同一睡姿。
基於上述提及睡姿與醫療應用的關聯性，如能開發出一套即時且自動化的睡姿辨識系統，時時監控需求病患的睡姿，當病患於床上出現不當的睡姿或是長期維持同一睡姿時，系統能對照護人員發出提醒，便對這些病患有莫大的幫助，也能使照護作業更有效率地進行。
本研究透過壓力感測床墊取得各式睡姿的壓力分佈數據，再以深度學習之卷積神經網路(Convolutional Neural Network, CNN)對數據進行訓練並建模，進而開發出以Android行動裝置為終端運算裝置的自動化即時睡姿辨識系統。本研究分為以下三個步驟：1.定義睡姿與數據收集；2.以卷積神經網路建模；3.開發Android應用程式進行睡姿辨識。本研究使用的壓力分佈數據是透過10名受試者根據自行定義之睡姿躺在壓力感測床墊上收集而來。經過卷積神經網路建模後，本研究將以LOSO(Leave-One-Subject-Out)的方式來驗證模型準確率。透過Android行動裝置作為運算端與介面端的系統架構，其優點為低時間延遲、靈活開發性與高機動性。

Sleep posture is not only one of the important factors in sleep quality assessment, but also associated with some chronic diseases. In medical condition, sleeping posture has a considerable impact on obstructive sleep apnea. Patients with sleep apnea should avoid sleep in the supine position and try to sleep with lateral postures. In addition to sleep apnea, sleep posture is also related to pressure ulcer prevention. Pressure ulcers are wounds of the skin and deeper soft tissue that occur in areas of bony prominence. Patients who have been in bed for a long time after surgery and mobility problems are unable to turn over on their own, which can easily cause pressure ulcers. These patients need caregivers to help them turn over regularly to avoid maintaining the same sleep posture for a long time.
Based on the relevance of the aforementioned sleep posture to medical applications, if a set of real-time automatic sleep posture recognition system can be developed to monitor the sleep posture of patients in need when they appear inappropriate sleep posture or maintain the same sleeping position for a long time, the system can send out the nursing staffs. As a reminder, it can greatly help these patients and make care operations more efficient.
In this study, we used a pressure-sensitive mattress to obtain pressure distribution data for various sleep postures, then we used CNN (convolutional neural network) to train the data and get the model. Finally, we developed a real-time automatic sleep posture recognition system using an Android mobile device as a terminal computing device. This research can be divided into the following three steps: 1. Define the sleep postures and data collection; 2. Model by CNN; 3. Develop an Android application to identify the sleep posture. The pressure distribution data used in this study was collected by 10 subjects lying on a pressure-sensitive mattress according to the defined sleep postures. After modeling by CNN, we will validate the model accuracy by using the Leave-One-Subject-Out (LOSO) method. The Android mobile device is used as the computing terminal and the interface terminal, the advantages of this system structure are low time latency, flexible development and high mobility.

[1] A. Oksenberg, I. Khamaysi, D. S. Silverberg, and A. Tarasiuk, “Association of body position with severity of apneic events in patients with severe nonpositional obstructive sleep apnea,” Chest, vol. 118, no. 4, pp. 1018– 1024, 2000.
[2] J. H. Walsh, M. S. Leigh, A. Paduch, K. J. Maddison, D. L. Philippe, J. J. Armstrong, D. D. Sampson, D. R. Hillman, and P. R. Eastwood, “Effect of body posture on pharyngeal shape and size in adults with and without obstructive sleep apnea,” Sleep, vol. 31, pp. 1543–1549, 2008.
[3] D. J. Margolis, J. Knauss, W. Bilker, and M. Baumgarten, "Medical conditions as risk factors for pressure ulcers in an outpatient setting," Age Ageing, vol. 32, pp. 259-264, 2003.
[4] G. Matar and J. Lina, “Unobtrusive sleep monitoring using cardiac, breathing and movements activities: an exhaustive review,” IEEE Access, vol. PP, no. c, pp. 1–1, 2018.
[5] A. Sadeh and C. Acebo, “The role of actigraphy in sleep medicine,” Sleep Med. Rev., vol. 6, no. 2, pp. 113–124, 2002.
[6] Y. Kishimoto, A. Akahori, and K. Oguri, “Estimation of sleeping posture for m-health by a wearable tri-axis accelerometer,” in Medical Devices and Biosensors. 3rd IEEE/EMBS International Summer School, Sept. 2006, pp. 45–48.
[7] K. M. Chang and S. H. Liu, “Wireless portable electrocardiogram and a tri-axis accelerometer implementation and application on sleep activity monitoring,” Telemed. E-Health, vol. 17, pp. 177–184, 2011.
[8] H. Yoon, S. Hwang, D. Jung, S. Choi, K. Joo, J. Choi, Y. Lee, D.-U. Jeong and K. Park “Estimation of sleep posture using a patch-type accelerometer based device,” in Proc. IEEE 37th Annu. Int. Conf. Eng. Med. Biol. Soc., pp. 4942–4945, 2015.
[9] K. Nakajima, Y. Matsumoto and T. Tamura, “A monitor for posture changes and respiration in bed using real time image sequence analysis,” in Engineering in Medicine and Biology Society, vol. 1, pp. 51–54, 2000.
[10] W.-H. Liao and C.-M. Yang, “Video-based activity and movement pattern analysis in overnight sleep studies,” in Pattern Recognition, 2008. ICPR 2008. 19th International Conference on, pp. 1–4, 2008.
[11] T. Grimm, M. Martinez, A. Benz and R. Stiefelhagen, “Sleep position classification from a depth camera using bed aligned maps,” International Conference on Pattern Recognition, 2016.
[12] F. Deng, J. Dong, X. Wang, Y. Fang, Y. Liu, Z. Yu, J. Liu, and F. Chen, “Design and implementation of a noncontact sleep monitoring system using infrared cameras and motion sensor,” IEEE Transactions on Instrumentation and Measurement, vol. 67, no. 7, pp. 1555–1563, 2018.
[13] S. M. Mohammadi, M. Alnowami, S. Khan, D. J. Dijk, A. Hilton, and K. Wells, “Sleep posture classification using a convolutional neural network,” in 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE, 2018, pp. 1–4.
[14] J. J. Liu, W. Xu, M.-C. Huang, N. Alshurafa, M. Sarrafzadeh, N. Raut and B. Yadegar, “Sleep posture analysis using a dense pressure sensitive bedsheet,” Pervasive Mobile Comput. J., vol. 10, no. 2, pp. 34–50, 2014.
[15] R.-S. Hsiao, T.‑X. Chen, M. A. Bitew, C.‑H. Kao and T.‑Y. Li, "Sleeping posture recognition using fuzzy c-means algorithm." Biomedical engineering online, vol. 17, no.2, pp.1-19, 2018.
[16] X. Xu, F. Lin, A. Wang, Y. Hu, M. C. Huang, and W. Xu, ‘‘Body-earth mover’s distance: A matching-based approach for sleep posture recognition,’’ IEEE Trans. Biomed. Circuits Syst., vol. 10, no. 5, pp. 1023–1035, 2016.
[17] G. Matar, J.-M. Lina, J. Carrier, A. Riley, and G. Kaddoum, ‘‘Internet of Things in sleep monitoring: An application for posture recognition using supervised learning,’’ in Proc. 18th Int. Conf. e-Health Netw., Appl. Services, pp. 1–6, 2016.
[18] Z. Liu, X. Wang, M. Su, and K. Lu, “A Method to Recognize Sleeping Position Using an CNN Model Based on Human Body Pressure Image.”2019 IEEE International Conference on Power, Intelligent Computing and Systems (ICPICS), pp.219-224, 2019.
[19] C. Idzikowski, “Learn to Sleep Well”, Watkins, 2010
[20] Sensor Products, “Ultra Thin Flexible Pressure Sensor”, [Online], Available: https://www.pibiox.com/webls-zh-tw/ultra-thin-pressure-sensor.html.
[21] 林大貴, “TensorFlow+Keras深度學習人工智慧實務應用”, 博碩出版社, 臺灣, 2018.
[22] K. Fukushima and S. Miyake, ‘‘Neocognitron: A self-organizing neural network model for a mechanism of visual pattern recognition,’’ in Competition and Cooperation in Neural Nets, Springer, pp. 267–285, 1982.
[23] T. Huang, ‘‘卷積神經網路—CNN運算流程’’, Medium [Online], Available: https://medium.com/@chih.sheng.huang821/
[24] M. Abadi, P. Barham, J. Chen, Z. Chen, A. Davis, J. Dean, M. Devin, S. Ghemawat, G. Irving, M. Isard, M. Kudlur, J. Levenberg, R. Monga, S. Moore, D. G. Murray, B. Steiner, P. Tucker, V. Vasudevan, P. Warden, M. Wicke, Y. Yu, and X. Zheng, “Tensorflow: A system for large-scale machine learning”, Tech. rep., Google Brain, 2016.
[25] L. Moroney, “Using TensorFlow Lite on Android”, [Online], Available:https://medium.com/tensorflow/using-tensorflow-lite-on-android-9bbc9cb7d69d.
[26] F. Chollet et al., “Keras: Deep Learning for Human”, GitHub [Online], Available: https://github.com/keras-team/keras.
[27] Y. LeCun, B. Boser, J. S. Denker, D. Henderson, R. E. Howard, W. Hubbard, and L. D. Jackel, “Backpropagation applied to handwritten zip code recognition,” Neural Computation, vol. 1, no. 4, pp. 541–551, 1989.