現今在臨床醫學上，判讀一位受試者是否具有身體的衰弱傾向時，通常都是透過問卷、量表、簡易身體數據等等來進行鑑定，而在不同問卷，其評分出來的標準皆為該問卷的主觀判讀，鮮少有相對應的實際身體數值於受試者身體狀況上，目前針對於此塊的研究領域仍少人涉略。
在本篇研究上，與專業護理人員配合，針對臺灣社區與醫院院區進行年長者衰弱資料蒐集，透過年長者之三軸加速度資訊進行訊號分析，提取各項特徵參數，將數據與特徵參數進行支援向量機進行人群分類學習，其分類標準以問卷篩檢與量表答案為主，將受試者依照標準分類，使支援向量機學習如何進行衰弱分類篩檢，達到問卷量表與身體數據的相互對應，使檢測結果可以更加準確。
透過以不同問卷作為分類標準，能夠有效使機器學習在判讀特徵時，能夠更完整性的妥善分類受試者，不同的問卷在判斷受試者狀況時，都會有獨特的分類依據，透過這些來補足受試者在判定衰弱時的正確定位。

Nowadays, in the clinical medicine, we usually use questionnaire, scale or simple body data to interpret the subject has frailty tendency or not. In different questionnaires, its standards are come from its subjective interpretation. There are few questionnaires which can correspond the subject`s body situation. Currently, few people are involved in the research area of this area.
In this search, we cooperate with professional nursing staff. We collect information on the elders` frailty in Taiwan`s communities and hospitals. Extract each feature parameter in triaxial acceleration by signal analysis. Putting data and feature parameters into the support vector machine to learn crowd classification. The classification criteria are based on questionnaire screening and scale answers. Classify the subjects according to the standard, and enable the support vector machine to learn how to conduct the classification screening for frailty. It achieves the correspondence between the questionnaire scale and the body data. It will make the test results can be more accurate.
By using different questionnaires as classification criteria, the machine learning can effectively classify the subjects more completely when interpreting the features. Different questionnaires will have unique classification basis when judging the condition of the subjects. Complement the correct positioning of the subjects when determining frailty.

摘要 i
Abstract ii
目錄 iii
表目錄 v
圖目錄 vii
一、介紹 1
1.1研究背景 1
1.2相關研究 1
1.3研究動機及目的 2
1.4問題描述和重要性 2
二、材料 3
2.1機器學習概述 3
2.1.1 KNN演算法（K nearest neighbor ，近鄰演算法） 3
2.1.2 SVM（Support vector machine，支援向量機 ） 3
2.2步態概述 3
2.3衰弱概述 4
2.4實驗器具、材料 6
2.4.1自製三軸加速度感測模組 7
2.5實驗相關問卷、指標、工具 9
2.5.1自擬式結構問卷 10
2.5.2察爾森共病症指標（Charlson Comorbidity Index，CCI） 10
2.5.3營養不良篩檢工具（Malnutrition Universal Screening Tool，MUST） 11
2.5.4日常生活功能量表（Activities of Daily Living，ADL） 12
2.5.5跌倒功效量表（Falls Efficacy Scale，FES） 12
2.5.6臨床衰弱量表（Clinical Frailty Scale，CFS） 12
2.5.7簡易認知功能評估量表（Short Portable Mental Status Questionnaire，SPMSQ） 13
2.5.8高齡者衰弱量表（Kihon Checklist，KCL） 14
2.6實驗相關測驗方式 14
2.6.1計時起走測試（Time up and Go，TUG） 14
2.6.2十公尺行走測試（Timed 10-Meter Walk Test，10MWT） 15
2.7檢測流程 15
三、方法 18
3.1訊號前置處理 18
3.2訊號計算處理 20
3.3訊號特徵設計、提取方法 22
3.3.1平均速率 23
3.3.2平均步頻 23
3.3.3平均步伐站立百分比 23
3.3.4平均步伐擺盪百分比 24
3.3.5左右腳步伐相關性 24
3.4機器學習模型 27
四、結果 28
4.1受試者 28
4.2問卷最佳特徵篩檢 29
4.3步態最佳特徵篩檢 32
4.4綜合特徵 34
五、討論 37
參考文獻 38

[1]黄菩臣, 練作為, 陶敏, 楊永強, 杜江, and 趙蘊龍, "基於MIMIC-III 資料庫對 ICU 患者结局預測的研究," 中國醫療設備, vol. 34, no. 12, pp. 92-96, 2019.
[2]H. Uchida et al., "The related factors of frailty risk in Japanese elderly women," in World Automation Congress 2012, 2012, pp. 1-5: IEEE.
[3]A. Vercelli et al., "My-active and healthy ageing (My-AHA): An ICT platform to detect frailty risk and propose intervention," in 2017 25th International Conference on Software, Telecommunications and Computer Networks (SoftCOM), 2017, pp. 1-4: IEEE.
[4]L. P. Fried et al., "Frailty in older adults: Evidence for a phenotype," Journals of Gerontology Series a-Biological Sciences and Medical Sciences, vol. 56, no. 3, pp. M146-M156, Mar 2001.
[5]李棟洲 and 武俊傑, "老年步態障礙," 台灣老年醫學暨老年學雜誌, vol. 10, no. 1, pp. 1-15, 2015.
[6]M. E. Charlson, P. Pompei, K. L. Ales, and C. R. J. J. o. C. E. MacKenzie, "A new method of classifying prognostic comorbidity in longitudinal studies: development and validation," vol. 40, no. 5, pp. 373-383, 1987.
[7]R. J. Stratton et al., "Malnutrition in hospital outpatients and inpatients: prevalence, concurrent validity and ease of use of the ‘malnutrition universal screening tool’(‘MUST’) for adults," vol. 92, no. 5, pp. 799-808, 2004.
[8]J. N. Morris, B. E. Fries, and S. A. J. T. J. o. G. S. A. Morris, "Scaling ADLs within the MDS," vol. 54, no. 11, pp. M546-M553, 1999.
[9]L. Yardley et al., "Development and initial validation of the Falls Efficacy Scale-International (FES-I)," vol. 34, no. 6, pp. 614-619, 2005.
[10]K. Rockwood et al., "A global clinical measure of fitness and frailty in elderly people," vol. 173, no. 5, pp. 489-495, 2005.
[11]E. J. J. o. t. A. G. S. Pfeiffer, "A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients," vol. 23, no. 10, pp. 433-441, 1975.
[12]D. Podsiadlo and S. J. J. o. t. A. g. S. Richardson, "The timed “Up & Go”: a test of basic functional mobility for frail elderly persons," vol. 39, no. 2, pp. 142-148, 1991.
[13]R. W. J. J. o. g. p. t. Bohannon, "Reference values for the timed up and go test: a descriptive meta-analysis," vol. 29, no. 2, pp. 64-68, 2006.
[14]S. L. Wolf, P. A. Catlin, K. Gage, K. Gurucharri, R. Robertson, and K. J. P. T. Stephen, "Establishing the reliability and validity of measurements of walking time using the Emory Functional Ambulation Profile," vol. 79, no. 12, pp. 1122-1133, 1999.
[15]R. C. Isles, N. L. L. Choy, M. Steer, and J. C. J. J. o. t. A. G. S. Nitz, "Normal values of balance tests in women aged 20–80," vol. 52, no. 8, pp. 1367-1372, 2004.
[16]王玥心, 張曉婷, 林明慧, 陳曾基, and 黃信彰, "老人衰弱症的評估與預後," 家庭醫學與基層醫療, vol. 34, no. 9, pp. 252-257, 2019