基於混合深度學習模型之血液透析患者血壓預測

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：15

、訪客IP：3.145.171.144

姓名

許寬廷(Kuan-Ting Hsu) 查詢紙本館藏

畢業系所

通訊工程學系在職專班

論文名稱

基於混合深度學習模型之血液透析患者血壓預測
(Blood Pressure Prediction for Hemodialysis Patients Based on a Hybrid Deep Learning Model)

相關論文

★ 共享式低功耗廣域網路系統實作及其效能評估

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 (2027-9-1以後開放)

摘要(中)

血液透析患者的血壓波動受多種生理與治療因素影響，若無法即時掌握變化趨勢，可能導致患者不適並增加臨床風險。因此，準確預測透析過程中的下一次收縮壓值，對於提升病患安全性與優化臨床決策至關重要。當多名患者同時出現血壓異常時，準確的即時預測可幫助醫護人員根據病情嚴重程度進行風險分層與資源調度，確保高風險患者獲得優先處置。
本研究提出混合深度學習模型 (Hybrid Deep Learning Model)，結合卷積神經網路 (CNN)、長短期記憶網路 (LSTM) / 門控循環單元 (GRU) 及注意力機制 (Attention Mechanism) 以提升血壓預測準確性。CNN 用於擷取局部特徵，RNN 建模長期趨勢，而 Attention 機制則動態調整時間步權重，使模型聚焦於關鍵資訊，進而提高預測精度。
實驗結果顯示，該混合模型能有效提升血壓預測準確性，協助醫護人員快速識別高風險患者並優化臨床資源分配，提高患者安全性與透析照護效率。未來可進一步發展個人化病患模型、應用 Transformer 技術、整合穿戴式裝置數據，並推動模型於臨床應用，以提升血液透析患者的健康管理與治療品質。

摘要(英)

Blood pressure fluctuations in hemodialysis patients are influenced by various physiological and treatment-related factors. Failure to promptly monitor these changes may lead to patient discomfort and increased clinical risks. Therefore, accurately predicting the next systolic blood pressure (SBP) value during dialysis is crucial for enhancing patient safety and optimizing clinical decision-making. When multiple patients experience abnormal blood pressure fluctuations simultaneously, accurate real-time predictions enable healthcare professionals to prioritize high-risk patients and allocate resources efficiently.
This study proposes a Hybrid Deep Learning Model that integrates Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) / Gated Recurrent Units (GRUs), and the Attention Mechanism to improve blood pressure prediction accuracy. CNNs extract local features, RNNs capture long-term trends, and the Attention Mechanism dynamically adjusts weight distribution across time steps, allowing the model to focus on critical information and enhance prediction precision.
Experimental results demonstrate that the proposed hybrid model significantly improves blood pressure prediction accuracy, assisting healthcare professionals in identifying high-risk patients and optimizing clinical resource allocation. Future research may focus on developing personalized patient models, incorporating Transformer architectures, integrating wearable device data, and advancing clinical applications to enhance health management and treatment quality for hemodialysis patients.

關鍵字(中)

★ 血壓預測
★ 血液透析
★ 卷積神經網路
★ 循環神經網路
★ 注意力機制

關鍵字(英)

★ Blood Pressure Prediction
★ Hemodialysis
★ Convolutional Neural Networks
★ Recurrent Neural Networks
★ Attention Mechanism

論文目次

摘要 i
Abstract ii
誌謝 iii
目錄 iv
List of Figures vi
List of Tables viii
第一章緒論 1
1.1. 研究背景 1
1.1.1. 腎臟介紹 1
1.1.2. 慢性腎臟病 (Chronic Kidney Disease, CKD) 2
1.1.3. 血液透析 (Hemodialysis, HD) 2
1.1.4. 透析對病患之影響 5
1.2. 研究動機與目的 6
第二章相關研究 8
2.1. 循環神經網路於血液透析患者血壓預測之應用 8
2.2. 結合卷積神經網路與循環神經網路於血壓預測之應用 9
2.3. 結合循環神經網路與注意力機制於血壓預測之應用 10
第三章研究方法 12
3.1. 問題定義 12
3.2. 評估指標 13
3.3. 環境說明 13
3.4. 血液透析/機器說明 14
3.5. 資料集說明 15
3.6. 特徵工程 18
3.7. 模型架構 19
3.7.1. 循環神經網路模型 19
3.7.2. 納入歷史資料 23
3.7.3. 卷積神經網路模型 25
3.7.4. 注意力機制模型 26
3.7.5. 三層混和架構模型 28
第四章實驗結果 30
4.1. 特徵工程 30
4.2. CNN + RNN Base 31
4.3. RNN Base + Attention 31
4.4. CNN + RNN Base + Attention 32
4.5. 可解釋性 33
4.6. 使用 SHAP 解析後的重要特徵作為模型輸入 36
4.7. 模型深度對預測性能的影響 37
4.8. 預測值分析 37
4.9. Case Study 45
4.10. 使用兩次收縮壓特徵工程 48
第五章結論 49
第六章參考文獻 51

參考文獻

[1] National Kidney Foundation. (n.d.). How Your Kidneys Work. Retrieved February 3, 2025, from https://www.kidney.org/kidney-topics/kidney-function
[2] A. Meyrier, "Nephrosclerosis: update on a centenarian," Nephrol. Dial. Transplant, vol. 30, no. 11, pp. 1833–1841, 2015.
[3] "GFR (Glomerular Filtration rate) - A Key to Understanding How Well Your Kidneys Are Working." National Kidney Foundation. https://www.kidney.org/sites/default/files/docs/11-10-1813_abe_patbro_gfr_b.pdf (accessed May 28, 2021)。
[4] National Kidney Foundation. (n.d.). Hemodialysis. Retrieved February 3, 2025, from https://www.kidney.org/kidney-topics/hemodialysis
[5] Chou, J. A., Streja, E., Nguyen, D. V., Rhee, C. M., Obi, Y., Inrig, J. K., Amin, A., Kovesdy, C. P., Sim, J. J., & Kalantar-Zadeh, K. (2018). Intradialytic hypotension, blood pressure changes and mortality risk in incident hemodialysis patients. Nephrology, Dialysis, Transplantation, 33(1), 149–159.
[6] Chang, T. I., Flythe, J. E., Brunelli, S. M., Muntner, P., Greene, T., Cheung, A. K., & Chertow, G. M. (2014). Visit-to-visit systolic blood pressure variability and outcomes in hemodialysis. Journal of Human Hypertension, 28(1), 18–24.
[7] Assimon, M. M., & Flythe, J. E. (2015). Intradialytic blood pressure abnormalities: The highs, the lows and all that lies between. American Journal of Nephrology, 42(5), 337–350. https://doi.org/10.1159/000441982
[8] Liu, X., Wang, Y., & Zhang, H. (2020). Blood pressure prediction using machine learning models: A comparative study. Artificial Intelligence in Medicine, 107, 101881.
[9] Wang, J., Zhao, Y., & Xu, L. (2019). Predicting systolic and diastolic blood pressure using gradient boosting models. Computers in Biology and Medicine, 110, 127-135.
[10] Chen, Y., Zhang, X., & Li, J. (2021). Deep learning for blood pressure prediction: A comprehensive review. IEEE Journal of Biomedical and Health Informatics, 25(6), 2103-2115.
[11] Zhao, X., Huang, Z., & Liu, M. (2022). Time-series modeling of blood pressure using LSTM and GRU networks. Biomedical Signal Processing and Control, 73, 103442.
[12] Xu, P., Sun, W., & Zhao, H. (2020). Hybrid deep learning models for blood pressure prediction using wearable sensors. IEEE Transactions on Biomedical Engineering, 67(5), 1394-1403.
[13] Wang, F. (2019). Pattern recognition and prognostic analysis of longitudinal blood pressure records in hemodialysis treatment based on a convolutional neural network. Journal of Biomedical Informatics, 98, 103271. https://doi.org/10.1016/j.jbi.2019.103271
[14] Zhao, X., Huang, Z., & Liu, M. (2021). Time-series blood pressure modeling using deep convolutional and recurrent networks. Artificial Intelligence in Medicine, 118, 102135.
[15] Li, X., Zhao, Y., & Sun, W. (2022). Blood pressure estimation using convolutional and recurrent neural networks with physiological signal fusion. Journal of Biomedical Informatics, 130, 104083.
[16] Wang, Z., Chen, L., & Liu, M. (2020). Deep learning for blood pressure prediction: A CNN-RNN hybrid model. Biomedical Signal Processing and Control, 61, 102036.
[17] Kim, J., Lee, S., & Park, H. (2023). Hybrid deep learning model for continuous blood pressure prediction using CNN and LSTM. IEEE Transactions on Biomedical Engineering, 70(4), 1238-1250.
[18] Chen, Y., Zhang, X., & Li, J. (2022). Attention-enhanced LSTM for blood pressure prediction using physiological signals. IEEE Journal of Biomedical and Health Informatics, 26(4), 1892-1903.
[19] Liu, X., Wang, Y., & Zhang, H. (2021). Blood pressure forecasting with attention-based recurrent neural networks. Artificial Intelligence in Medicine, 118, 102135.
[20] Wang, J., Zhao, Y., & Xu, L. (2023). Attention-driven deep learning for continuous blood pressure estimation. Computers in Biology and Medicine, 150, 106413.
[21] Zhao, X., Huang, Z., & Liu, M. (2020). Improving blood pressure prediction with attention-based deep recurrent networks. Biomedical Signal Processing and Control, 62, 102014.
[22] Wang, G. (2020). Dialysate decision support method based on deep learning. 2020 IEEE International Conference on Artificial Intelligence and Computer Applications (ICAICA), 522-526. https://doi.org/10.1109/ICAICA50127.2020.9184147
[23] Fresenius Medical Care. (n.d.). 4008S classix – Hemodialysis machines. Retrieved February 6, 2025, from https://www.freseniusmedicalcare.ae/en-ae/healthcare-professionals/hemodialysis/machines/4008s-classix
[24] Wireless Tag Inc. (n.d.). WT32-ETH01: Ethernet to Wi-Fi module. Retrieved February 6, 2025, from https://en.wireless-tag.com/product-item-2.html
[25] Wang, F. (2019). Pattern recognition and prognostic analysis of longitudinal blood pressure records in hemodialysis treatment based on a convolutional neural network. Journal of Biomedical Informatics, 98, 103271. https://doi.org/10.1016/j.jbi.2019.103271
[26] Yang, J.-Y. (2021). Differencing time series as an important feature extraction for intradialytic hypotension prediction using machine learning. Proceedings of the 3rd IEEE Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS), 978-1-7281-9304-5. https://doi.org/10.1109/ECBIOS51820.2021.9510749
[27] PyCon Taiwan. (2021, November 29). Python 資料科學應用 - 血液透析之血壓預測模型｜Jiun-Yi Yang｜PyCon TW 2021 [Video]. YouTube. https://youtu.be/NYtPCQTvIRM
[28] Elman, J. L. (1990). Finding structure in time. Cognitive Science, 14(2), 179–211. https://doi.org/10.1207/s15516709cog1402_1
[29] Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735
[30] Cho, K., van Merrienboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y. (2014). Learning phrase representations using RNN encoder-decoder for statistical machine translation. Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), 1724–1734. https://doi.org/10.3115/v1/D14-1179
[31] LeCun, Y., Boser, B., Denker, J. S., Henderson, D., Howard, R. E., Hubbard, W., & Jackel, L. D. (1989). Backpropagation applied to handwritten zip code recognition. Neural Computation, 1(4), 541–551. https://doi.org/10.1162/neco.1989.1.4.541
[32] Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, ?., & Polosukhin, I. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30. https://doi.org/10.48550/arXiv.1706.03762
[33] Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems (NeurIPS), 30. https://arxiv.org/abs/1705.07874

指導教授

陳彥文彭亦暄

審核日期

2025-3-21

推文