睡眠佔據了人類生命的約三分之一，對生理和心理健康有深遠的影響。準確的睡眠階段分類對於改善睡眠品質至關重要。雖然多導睡眠圖（PSG）被視為睡眠測量的黃金標準，但其複雜性和高成本限制了其應用。消費性穿戴式裝置透過提供連續、非侵入性的生理監測，提供了一種更易於實行的替代方案。
本論文提出了一種適用於低功耗穿戴設備的節能睡眠階層分類方法。我們的方法利用深度學習模型，通過減少 CNN 層深度、引入靈活的分層結構以及在多個任務中共享參數，以在效率和準確性之間達到最佳平衡來降低計算成本。
該模型採用了多階層分類架構，使用簡單分類器進行初步預測，並透過信心驅動的分層方法，使用深度分類器處理更複雜的案例。在兩個公開數據集上的實驗顯示，計算成本和模型參數顯著減少至原有的80%與90%，同時性能僅略有下降。甚至在相似參數下，能降低50%的計算量，並維持或提高效能。這種分層設計展現了在成本效益與分類準確性之間取得平衡的潛力，特別適用於低功耗穿戴設備的應用場景。
本研究為高效且準確的睡眠階層分類做出了貢獻，支持穿戴式裝置進行持續的過夜監測，並促進更好的睡眠品質。

Sleep occupies about one-third of human life and has a significant impact on both physiological and psychological health. Accurate sleep stage classification is essential for improving sleep quality. While polysomnography (PSG) is the gold standard for sleep measurement, its complexity and high cost limit its accessibility. Consumer wearable devices offer a more practical alternative by providing continuous, non-invasive physiological monitoring.
This thesis proposes a power-efficient sleep stage classification method tailored for low-power wearable devices. The approach leverages a deep learning model designed to minimize computational costs by reducing CNN layer depth, incorporating a flexible hierarchical structure, and sharing parameters across tasks to optimize the balance between efficiency and accuracy.
The model employs a multi-level classification architecture, using a simple classifier for initial predictions and a confidence-based hierarchical approach for more complex cases. Experiments on two public datasets demonstrate significant reductions in computational costs and model parameters—by up to 80% and 90%—with only a slight decrease in performance. Under similar parameter settings, the computational load can be reduced by 50% while maintaining or enhancing performance. This hierarchical design has the potential to improve both cost efficiency and classification accuracy.
This research advances efficient and accurate sleep stage classification, supporting the use of wearable devices for continuous overnight monitoring and improved sleep quality.

[1] R. Stickgold and M. P. Walker, “Sleep-dependent memory triage: evolving generalization through selective processing,” Nature Neuroscience, vol. 16, pp. 139–145, 2013.
[2] M. P. Walker and R. Stickgold, “Sleep, memory, and plasticity.” Annual review of psychology, vol. 57, pp. 139–66, 2006.
[3] L. Xie, H. Kang, Q. Xu, M. J. Chen, Y. Liao, M. Thiyagarajan, J. O’Donnell, D. J. R. Christensen, C. Nicholson, J. J. Iliff, T. Takano, R. Deane, and M. Nedergaard, “Sleep drives metabolite clearance from the adult brain,” Science, vol. 342, pp. 373 – 377, 2013.
[4] M. R. Irwin and M. R. Opp, “Sleep health: Reciprocal regulation of sleep and innate immunity,” Neuropsychopharmacology, vol. 42, pp. 129–155, 2017.
[5] M. P. Walker, “The role of sleep in cognition and emotion,” Annals of the New York Academy of Sciences, vol. 1156, 2009.
[6] D. J. Buysse, C. F. Reynolds, T. H. Monk, S. R. Berman, and D. J. Kupfer, “The pittsburgh sleep quality index: A new instrument for psychiatric practice and research,” Psychiatry Research, vol. 28, pp. 193–213, 1989.
[7] C. Iber, S. Ancoli-Israel, A. L. Chesson, and S. F. Quan, “The american academy of sleep medicine (aasm) manual for the scoring of sleep and associated events: Rules, terminology and technical specifications,” 2007.
[8] F. Laricchia, “2014-2028 total wearable shipments worldwide.” [Online]. Available: https://www.statista.com/statistics/437871/wearables-worldwide-shipments/
[9] R. Dobson, M. Stowell, J. Warren, T. Tane, L. Ni, Y. Gu, J. McCool, and R. Whittaker, “Use of consumer wearables in health research: Issues and considerations,” Journal of Medical Internet Research, vol. 25, 2023.
[10] M. R. Lujan, I. Perez-Pozuelo, and M. A. Grandner, “Past, present, and future of multisensory wearable technology to monitor sleep and circadian rhythms,” Frontiers in Digital Health, vol. 3, 2021.
[11] A. Henriksen, M. H. Mikalsen, A. Z. Woldaregay, M. Muzny, G. Hartvigsen, L. A. Hopstock, and S. Grimsgaard, “Using fitness trackers and smartwatches to measure physical activity in research: Analysis of consumer wrist-worn wearables,” Journal of Medical Internet Research, vol. 20, 2018.
[12] X. Liu and A. G. Richardson, “A system-on-chip for closed-loop optogenetic sleep modulation,” 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), pp. 5678–5681, 2021.
[13] A. Nguyen, G. Pogoncheff, B. X. Dong, N. Bui, H. Truong, N. Pham, L. T. Nguyen, H. H. Nguyen, S. Duong-Quy, S. Ha, and T. Vu, “A large-scale study of a sleep tracking and improving device with closed-loop and personalized real-time acoustic stimulation,” ArXiv, vol. abs/2211.02592, 2022.
[14] M. Olsen, J. M. Zeitzer, R. N. Richardson, P. Davidenko, P. J. Jennum, H. B. D. Sørensen, and E. J.-M. Mignot, “A flexible deep learning architecture for temporal sleep stage classification using accelerometry and photoplethysmography,” IEEE Transactions on Biomedical Engineering, vol. 70, pp. 228–237, 2022.
[15] X. Zhang, X. Zhou, M. Lin, and J. Sun, “Shufflenet: An extremely efficient convolutional neural network for mobile devices,” 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 6848–6856, 2017.
[16] Z. Yao and X. Liu, “A cnn-transformer deep learning model for real-time sleep stage classification in an energy-constrained wireless device*,” 2023 11th International IEEE/EMBS Conference on Neural Engineering (NER), pp. 1–4, 2022.
[17] A. Supratak, H. Dong, C. Wu, and Y. Guo, “Deepsleepnet: A model for automatic sleep stage scoring based on raw single-channel eeg,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 25, pp. 1998–2008, 2017.
[18] Y. Liao, C. Zhang, M. Zhang, Z. Wang, and X. Xie, “Lightsleepnet: Design of a personalized portable sleep staging system based on single-channel eeg,” IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 69, pp. 224–228, 2022.
[19] G. Liu, G. Wei, S. Sun, D. Mao, J. Zhang, D. Zhao, X.-L. Tian, X. Wang, and N. Chen, “Micro sleepnet: efficient deep learning model for mobile terminal real-time sleep staging,” Frontiers in Neuroscience, vol. 17, 2023.
[20] P. Fonseca, M. Ross, A. Cerny, P. Anderer, F. B. van Meulen, H. Janssen, A. Pijpers, S. Dujardin, P. V. van Hirtum, M. M. van Gilst, and S. Overeem, “A computationally efficient algorithm for wearable sleep staging in clinical populations,” Scientific Reports, vol. 13, 2023.
[21] G. Huang, Y. Yuan, G. Cao, and F. Ma, “Accsleepnet: An axis-aware hybrid deep fusion model for sleep stage classification using wrist-worn accelerometer data,” 2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 1005–1012, 2022.
[22] A. Sadeh, “The role and validity of actigraphy in sleep medicine: an update.” Sleep medicine reviews, vol. 15 4, pp. 259–67, 2011.