本研究設計了一套結合非侵入式設備與人工智慧的AI評分系統，解決居家骨盆底肌肉復健無法即時確認正確度及醫師無法追蹤病患情況的問題。透過濾波、切割優化及尺寸調整處理壓力感測數據，提升數據準確性，並根據病患肌力特性建立個人化評分模型。實驗顯示，儀器放置於大腿內側中央時，評分準確率較高（0.778）。
    利用邏輯迴歸與隨機森林分析，結果顯示，完成規律波形的病患復健效果較佳，且規律波形占比高於70%者，及未做腹部手術者，復健改善顯著。本論文設計AI評分系統，輔助居家骨盆底肌肉復健，並分析病患復健效果的關鍵因素。

This study designed an AI scoring system combining non-invasive devices and artificial intelligence to address the issues of not being able to immediately verify the accuracy of home-based pelvic floor muscle rehabilitation and the inability for physicians to track patients' conditions. By applying filtering, segmentation optimization, and size adjustment to pressure sensor data, the accuracy of the data was enhanced, and a personalized scoring model was established based on each patient's muscle strength characteristics. Experiments showed that placing the device at the center of the inner thigh resulted in a higher scoring accuracy (0.778).
Using logistic regression and random forest analysis, the results indicated that patients who completed regular waveforms had better rehabilitation outcomes. Additionally, patients with more than 70% regular waveform completion and those who had not undergone abdominal surgery showed significant improvements in rehabilitation. This paper designs an AI scoring system to assist in home-based pelvic floor muscle rehabilitation and analyzes the key factors influencing patients' rehabilitation outcomes.

目錄

第一章 研究背景與動機	1
第一節 研究背景	1
第二節 研究動機	2
第三節 研究目的	4
第四節 論文架構	5
第二章 文獻探討	6
第一節 骨盆底肌肉訓練	6
第二節 卷積濾波	7
第三節 邏輯迴歸	8
第四節 隨機森林	8
第五節 儀器歷屆版本	9
第三章 研究方法	11
第一節 資料收集	13
第二節 資料預處理	17
第三節 模型建置	26
第四節 問卷相關性	28
第四章 實作與問卷分析	31
第一節 實作	31
第二節 評分模型驗證	37
第三節 問卷分析	39
第五章 結論與未來展望	43
第一節 結論	43
第二節 未來展望	44
參考文獻	46
 
圖目錄

圖 2.1  骨盆底肌肉位置示意圖	7
圖 2.2  骨盆底肌肉運動儀器歷屆版本設計	10
圖 3.1  骨盆底肌肉運動復健系統的應用情境流程圖	11
圖 3.2  居家復健儀器操作與即時評分回饋流程圖	12
圖 3.3  研究流程圖	12
圖 3.4  運動數據收集問題與解決方案示意圖	14
圖 3.5  運動數據收集儀器結構與設計圖	15
圖 3.6  骨盆底肌肉訓練收縮與放鬆流程圖	15
圖 3.7  壓力數據記錄	16
圖 3.8  骨盆底肌肉收縮波形圖	16
圖 3.9  骨盆底肌肉收縮的波形特徵	17
圖 3.10 受試者資料衍生與新增項目	18
圖 3.11  混亂與連續波形示例	19
圖 3.12  波形持續時間不一致示例	19
圖 3.13　壓力波形所需解決的問題	20
圖 3.14　波形預處理流程圖	20
圖 3.15  雜訊示意圖	21
圖 3.16  波形不穩定示意圖	22
圖 3.17  移動平均濾波的雜訊過濾導致波形放大	22
圖 3.18  移動平均濾波的雜訊過濾效果	23
圖 3.19  單一波形變化	24
圖 3.20  曲線得分P的變化示意圖	25
圖 3.21  切割波形示意圖	25
圖 3.22  理想波形	27
圖 3.23  原始波形經過調整尺寸的變化	27
圖 3.24  評分模型示意圖	28
圖 3.25  波形評分模型運算結果	28
圖 3.26  問卷相關性因子分析流程圖	29
圖 4.1  波形抓取問題分類	32
圖 4.2  不同壓力類型下卷積尺寸評估	33
圖 4.3  不同版本切割模型設計及比較	34
圖 4.4  不同病患侵入式骨盆底肌肉復健的最高電壓值	35
圖 4.5  大腿夾住儀器時的壓力變化	35
圖 4.6  手扶住儀器時的壓力變化	36
圖 4.7  個人化評分模型建置流程	36
圖 4.8  儀器放置於不同位置的各類波形最大值盒鬚圖	38
圖 4.9  隨機森林結果	41
圖 4.10  基於分群規則的隨機森林結果整理	41
 
表目錄

表 1.1  非手術的尿失禁治療方式優缺點比較	3
表 2.1  骨盆底肌肉運動儀器各版本設計與功能比較	10
表 3.1  切割模型參數	24
表 3.2  應變數與自變數資料分群依據	30
表 4.1  濾除雜訊模型參數及方法	33
表 4.2  中央位置波形分類的混淆矩陣	37
表 4.3  接近膝蓋位置波形分類的混淆矩陣	37
表 4.4  問卷分數差逐步迴歸分析結果	39
表 4.5  問卷分數差與規律波形的邏輯迴歸結果	40
表 4.6  隨機森林浴測結果的混淆矩陣	42
表 5.1  改善程度顯著與不顯著之特徵分析	44

[1]	S. Batmani, R. Jalali, M. Mohammadi, and S. Bokaee, “Correction: Prevalence and factors related to urinary incontinence in older adults women worldwide: a comprehensive systematic review and meta-analysis of observational studies,” BMC Geriatr., vol. 22, p. 454, May 2022, doi: 10.1186/s12877-022-03111-6.
[2]	A. Alizadeh et al., “Prevalence and severity of urinary incontinence and associated factors in Iranian postmenopausal women: a cross-sectional study,” BMC Urol., vol. 23, p. 18, Feb. 2023, doi: 10.1186/s12894-023-01186-w.
[3]	A. Onishi and A. Shibata, “Prevalence and sociodemographic correlates of urinary incontinence in Japanese women: A web-based cross-sectional study,” Womens Health Lond. Engl., vol. 19, p. 17455057231207754, 2023, doi: 10.1177/17455057231207754.
[4]	B. Sumarsono et al., “The prevalence of urinary incontinence in men and women aged 40 years or over in China, Taiwan and South Korea: A cross-sectional, prevalence-based study,” LUTS Low. Urin. Tract Symptoms, vol. 12, no. 3, pp. 223–234, 2020, doi: 10.1111/luts.12308.
[5]	Q. Wang, Y.-Z. Que, X.-Y. Wan, and C.-Q. Lin, “Prevalence, Risk Factors, and Impact on Life of Female Urinary Incontinence: An Epidemiological Survey of 9584 Women in a Region of Southeastern China,” Risk Manag. Healthc. Policy, vol. 16, pp. 1477–1487, Aug. 2023, doi: 10.2147/RMHP.S421488.
[6]	L. C. G. Jensen and S. Axelsen, “Urinary incontinence and work capacity,” Dan. Med. J., vol. 70, no. 10, p. A04230271, Sep. 2023.
[7]	P.-M. Chow, Y.-C. Chuang, K. C. P. Hsu, Y.-C. Shen, and S.-P. Liu, “Impact of Female Stress Urinary Incontinence on Quality of Life, Mental Health, Work Limitation, and Healthcare Seeking in China, Taiwan, and South Korea (LUTS Asia): Results from a Cross-Sectional, Population-Based Study,” Int. J. Womens Health, vol. 14, pp. 1871–1880, Dec. 2022, doi: 10.2147/IJWH.S383651.
[8]	M. Kopańska, S. Torices, J. Czech, W. Koziara, M. Toborek, and Ł. Dobrek, “Urinary incontinence in women: biofeedback as an innovative treatment method,” Ther. Adv. Urol., vol. 12, p. 1756287220934359, 2020, doi: 10.1177/1756287220934359.
[9]	E. Erekson et al., “Outpatient Evaluation and Management Visits for Urinary Incontinence in Older Women,” J. Urol., vol. 202, no. 2, pp. 333–338, Aug. 2019, doi: 10.1097/JU.0000000000000223.
[10]	A. H. Kegel, “Progressive resistance exercise in the functional restoration of the perineal muscles,” Am. J. Obstet. Gynecol., vol. 56, no. 2, pp. 238–248, Aug. 1948, doi: 10.1016/0002-9378(48)90266-X.
[11]	A. H. Abu Raddaha and E. H. Nasr, “Kegel Exercise Training Program among Women with Urinary Incontinence,” Healthc. Basel Switz., vol. 10, no. 12, p. 2359, Nov. 2022, doi: 10.3390/healthcare10122359.
[12]	C. Dumoulin, L. P. Cacciari, and E. J. C. Hay‐Smith, “Pelvic floor muscle training versus no treatment, or inactive control treatments, for urinary incontinence in women,” Cochrane Database Syst. Rev., vol. 2018, no. 10, p. CD005654, Oct. 2018, doi: 10.1002/14651858.CD005654.pub4.
[13]	F. F. Fitz, L. M. Paladini, L. de A. Ferreira, M. M. Gimenez, M. A. T. Bortolini, and R. A. Castro, “Ability to contract the pelvic floor muscles and association with muscle function in incontinent women,” Int. Urogynecology J., vol. 31, no. 11, pp. 2337–2344, Nov. 2020, doi: 10.1007/s00192-020-04469-0.
[14]	J. Barbosa-Silva, S. C. Zanello, C. H. Jorge, and P. Driusso, “Do young women have an accurate perception about their pelvic floor muscle contraction? An agreement study about self-perception and physical evaluation of the pelvic muscles contraction by the PERFECT scheme,” J. Bodyw. Mov. Ther., vol. 40, pp. 411–416, Oct. 2024, doi: 10.1016/j.jbmt.2024.04.054.
[15]	Uechi N. et al., “Do women have an accurate perception of their pelvic floor muscle contraction? A cross‐sectional study”, doi: 10.1002/nau.24214.
[16]	“Using Adaptive Wireless Transmission of Wearable Sensor Device for Target Heart Rate Monitoring of Sports Information.” Accessed: Dec. 26, 2024. [Online]. Available: https://ieeexplore.ieee.org/document/9241836
[17]	N. R. Panda, “A Review on Logistic Regression in Medical Research,” Natl. J. Community Med., vol. 13, no. 04, Art. no. 04, Apr. 2022, doi: 10.55489/njcm.134202222.
[18]	J. Hu and S. Szymczak, “A review on longitudinal data analysis with random forest,” Brief. Bioinform., vol. 24, no. 2, p. bbad002, Mar. 2023, doi: 10.1093/bib/bbad002.
[19]	N. Kumagai and M. Jakovljević, “Random forest model used to predict the medical out-of-pocket costs of hypertensive patients,” Front. Public Health, vol. 12, Jul. 2024, doi: 10.3389/fpubh.2024.1382354.