研究生: |
謝和峰 Ho-Feng Hsieh |
---|---|
論文名稱: |
結合穿戴式脈波量測與分類分析應用於血管疾病風險早期評估與預測:代謝症候群與糖尿病前期 Combining wearable pulse wave measurement with classification analysis for early assessment and prediction of vascular disease risk: Metabolic Syndrome and Pre-Diabetes |
指導教授: |
許昕
Hsin Hsiu |
口試委員: |
許昕
吳立偉 林上智 |
學位類別: |
碩士 Master |
系所名稱: |
應用科技學院 - 醫學工程研究所 Graduate Institute of Biomedical Engineering |
論文出版年: | 2023 |
畢業學年度: | 112 |
語文別: | 中文 |
論文頁數: | 157 |
中文關鍵詞: | 代謝症候群 、糖尿病 、動脈硬化 、機器學習(ML) 、穿戴式脈波量測裝置 、脈波分佈區間分析法(PDA) 、循環疾病 |
外文關鍵詞: | metabolic syndrome, diabetes, arteriosclerosis, machine learning (ML), wearable pulse wave measurement device, pulse distribution interval analysis (PDA), circulatory diseases |
相關次數: | 點閱:92 下載:0 |
分享至: |
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
摘要
現代人因工作壓力及飲食習慣常常處於不良的生活型態,隨著年齡的老化,不斷累積血管中的脂肪與膽固醇,加上壓力造成內分泌失調,導致血糖上升,長年下來容易造成代謝症候群引發血管特性改變,進而增加現代人得到糖尿病、高血壓、中風等相關心血管疾病的風險,甚至誘發動脈粥狀硬化,也就是所謂的「動脈硬化」。目前能夠有效且檢測動脈硬化的儀器的成本通常都較為高昂且需要專業醫檢人員協助進行操作與量測,量測過程也較易受限於空間範圍大小,一般民眾通常都在健檢時都較傾向於簡易量測的方法,但這些方法除了需要一般民眾親自進入醫院接受侵入式檢測(如:抽血)、BMI等,同時需要等待一段時間才能獲取結果報告,因此,一套擁有便利性、即時性且具有高準確度的診斷工具,對於醫院及民眾是迫切且尚未滿足的醫療需求,希望能在各種環境即時測量,搭配AL將數據進行分析及應用,便能有效降低且預防上述疾病的發生機率。
目前已發現重要循環早期疾病(如:代謝症候群及糖尿病前期)皆可能影響動脈血管特性的變化。本研究將使用本實驗室自行開發的非侵入式穿戴式脈波量測系統應用於代謝症候群及糖尿病中,量測生理訊號包含:
1. 心電訊號描述放大器心電訊號描述放大器(Electrocardiography, ECG)
2. 動脈血壓波形感測器(Blood Pressure Waveform, BPW)
3. 血管光容積感測器(Photoplethysmography, PPG)
並與內湖三軍總醫院老人健檢中心合作,分別探討高齡化族群的代謝症候群(Metabolic Syndrome, MetS)與完全健康的民眾(Control)、糖尿病(Diabetes Mellitus, DM)與完全健康的民眾(Control)之間的血管特性差異,同時,也納入了代謝症候群前期(Pre-MetS)及糖尿病前期(Pre-DM)進行合併探討。
在收案過程中,接受雙盲實驗的民眾數據會進行頻域分析計算出生理參數,配合2種分析方式,分別為機器學習(ML)分析(以40個頻域諧波參數為特徵:能量比例(Cn)、變異係數(CVn)、相位角(Pn)及標準偏差(Pn_SD))以及脈波分佈區間分析法(PDA)來進行數據分析,並透過機器學習的8種演算法,建立出最合適的機器學習模型,將其應用在分類代謝症候群及糖尿病與前期的數據上。
本實驗結果發現,代謝症候群及糖尿病前期的主題皆具有顯著差異。基於這些分析,透過ML三倍交叉驗證(3-fold cross-validation)與PDA的表現,發現PDA與ML的效果皆有一定程度的分類能力且具有實際應用的潛力,後續再進行了其他干擾因子的分類分析。
1. 使用BPW及PPG量測,發現代謝症候群與糖尿病前期於生物統計上有顯著差異,因此有助於代謝症候群與糖尿病前期的早期預防。
2. 基於生物統計顯著差異,發現ML與PDA的分類效果有助於臨床研究早期進行AI分析的可行性評估。
可發現代謝症候群與糖尿病前期皆屬於早期循環疾病,並且在本穿戴式脈波量測裝置上產生顯著差異,代表直接疾病對血管特性帶來明顯的影響,可能是脈波分析在分類效果上可具有良好的成果原因,因此,本穿戴式脈波量測裝置除了擁有非侵入式、量測簡易快速且攜帶方便等優勢,配合ML及PDA的分類方式,在臨床層面上也有一定的應用價值,符合院方及民眾目前醫療市場需求主流,並且有助於加速未來醫療市場的發展及強化居家保健上的疾病預防措施。
Abstract
Modern people often have unhealthy lifestyles due to work pressure and dietary habits. As they age, fat and cholesterol accumulate in their blood vessels, and stress causes hormonal imbalances, leading to elevated blood sugar levels. Over time, this can result in metabolic syndrome, which causes changes in blood vessel characteristics and increases the risk of cardiovascular diseases such as diabetes, hypertension, and stroke. It can even lead to atherosclerosis, also known as "hardening of the arteries." Currently, the cost of effective instruments for detecting and measuring atherosclerosis is usually high, and they require assistance from professional medical personnel. The measurement process is also often limited by the size of the space available. Generally, people prefer simpler measurement methods during health check-ups. However, these methods require individuals to undergo invasive tests (such as blood tests) and measure their BMI, and they also need to wait for some time to receive the results. Therefore, there is an urgent and unmet medical need for a diagnostic tool that is convenient, real-time, and highly accurate. It is hoped that such a tool can be used to measure in various environments and analyze and apply the data with the help of AI, effectively reducing and preventing the occurrence of the aforementioned diseases.
Currently, early-stage diseases such as metabolic syndrome and prediabetes have been found to potentially affect changes in arterial vascular characteristics. In this study, our laboratory-developed non-invasive wearable pulse wave measurement system will be applied to metabolic syndrome and diabetes. The system measures physiological signals including:
1. Electrocardiography (ECG)
2. Blood Pressure Waveform (BPW)
3. Photoplethysmography (PPG)
We will collaborate with the Elderly Health Examination Center at Tri-Service General Hospital in Neihu to investigate the differences in vascular characteristics between the aging population with metabolic syndrome (MetS) and completely healthy individuals (Control), as well as between individuals with diabetes mellitus (DM) and completely healthy individuals (Control). Additionally, we will also include individuals with prediabetes (Pre-DM) and prediabetes metabolic syndrome (Pre-MetS) for combined analysis.
During the enrollment process, the data of participants undergoing double-blind experiments will undergo frequency domain analysis to calculate physiological parameters. This analysis is done using two methods: machine learning (ML) analysis, which uses 40 frequency domain harmonic parameters as features (energy ratio (Cn), coefficient of variation (CVn), phase angle (Pn), and standard deviation (Pn_SD)), and pulse distribution interval analysis (PDA). The data is then analyzed using eight machine learning algorithms to establish the most suitable machine learning model, which is then applied to classify data related to metabolic syndrome and prediabetes.
In the study, it was found that there were significant differences in the characteristics of metabolic syndrome and prediabetes. Based on the performance of ML and PDA through 3-fold cross-validation, it was discovered that both ML and PDA had a certain degree of classification ability and practical application potential. Subsequent classification analyses were conducted on other interfering factors.
Using BPW and PPG measurements, it was found that there were significant differences in the biostatistics of metabolic syndrome and prediabetes, which could contribute to early prevention of these conditions.
1. Based on the biological statistics, significant differences were found between metabolic syndrome and prediabetes, as measured by BPW and PPG. This finding can contribute to early prevention of metabolic syndrome and prediabetes.
2. Based on the significant differences in biological statistics, the classification effects of ML and PDA are helpful for the feasibility assessment of early AI analysis in clinical research.
It is found that both metabolic syndrome and prediabetes belong to early-stage circulatory diseases, and they show significant differences in the wearable pulse wave measurement device. This indicates that the diseases directly affect vascular characteristics. The good classification results of pulse wave analysis may be the reason for this. Therefore, the wearable pulse wave measurement device has advantages such as non-invasiveness, easy and fast measurement, and convenient portability. Combined with ML and PDA classification methods, it also has certain clinical application value. It meets the current mainstream demands of hospitals and the public in the medical market and helps accelerate the development of the future medical market and strengthen disease prevention measures in home healthcare.
[1] 胰島素阻抗是什麼?糖尿病前期又是什麼?用飲食、運動逆轉糖尿病
https://heho.com.tw/archives/20980
[2] 動脈硬化是什麼?造成動脈硬化的八大危險因子-智抗糖部落格
https://blog.health2sync.com/what-is-arteriosclerosis/
[3] CCMP96-TP-201 脈波信號與PPG信號特徵之相關性研究
https://dep.mohw.gov.tw/DOCMAP/cp-719-30696-108.html
[4] Dagliati A, Marini S, Sacchi L, Cogni G, Teliti M, Tibollo V, De Cata P, Chiovato L, Bellazzi R. Machine Learning Methods to Predict Diabetes Complications. J Diabetes Sci Technol. 2018 Mar;12(2):295-302. doi: 10.1177/1932296817706375. Epub 2017 May 12. PMID: 28494618; PMCID: PMC5851210.
[5] Krishnamoorthi R, Joshi S, Almarzouki HZ, Shukla PK, Rizwan A, Kalpana C, Tiwari B. A Novel Diabetes Healthcare Disease Prediction Framework Using Machine Learning Techniques. J Healthc Eng. 2022 Jan 11;2022:1684017. doi: 10.1155/2022/1684017. PMID: 35070225; PMCID: PMC8767376.
[6] Ellahham S. Artificial Intelligence: The Future for Diabetes Care. Am J Med. 2020 Aug;133(8):895-900. doi: 10.1016/j.amjmed.2020.03.033. Epub 2020 Apr 20. PMID: 32325045.
[7] Choi SB, Kim WJ, Yoo TK, Park JS, Chung JW, Lee YH, Kang ES, Kim DW. Screening for prediabetes using machine learning models. Comput Math Methods Med. 2014;2014:618976. doi: 10.1155/2014/618976. Epub 2014 Jul 16. PMID: 25165484; PMCID: PMC4140121.
[8] Birk N, Matsuzaki M, Fung TT, Li Y, Batis C, Stampfer MJ, Deitchler M, Willett WC, Fawzi WW, Bromage S, Kinra S, Bhupathiraju SN, Lake E. Exploration of Machine Learning and Statistical Techniques in Development of a Low-Cost Screening Method Featuring the Global Diet Quality Score for Detecting Prediabetes in Rural India. J Nutr. 2021 Oct 23;151(12 Suppl 2):110S-118S. doi: 10.1093/jn/nxab281. PMID: 34689190; PMCID: PMC8542097.
[9] 血壓能看出血管阻塞或硬化!6種血壓高低差看出血管病變
https://www.edh.tw/article/26012
[10] 肚子越大命越短」 腰圍超過這數字 糖尿病風險增10倍
https://www.chinatimes.com/realtimenews/20230406001300-260418?chdtv
[11] 口服抗糖尿病藥物用法及作用
http://www.tade.org.tw/view/web_pdf/%E5%8F%A3%E6%9C%8D%E9%99%8D%E8%A1%80%E7%B3%96%E8%97%A5%E7%94%A8%E6%B3%95%E5%8F%8A%E4%BD%9C%E7%94%A8.pdf
[12] 臨床性別醫學手冊:第二型糖尿病
https://gms.kmu.edu.tw/?p=981
[13] 「代謝症候群」是什麼?跟心血管疾病有什麼關係?
https://www.thenewslens.com/article/124436
[14] 代謝症候群,肥胖,心血管疾病,高血壓 - 存奕美學診所
https://drglowbeauty.com.tw/news/article/metabolic_syndrome
[15] 腹部肥胖恐增心血管風險醫:這1時間點的女性最危險!
https//www.commonhealth.com.tw/article/83842?rec=ch-to-other-i2i-1-visitor&from=ch&from_id=article%20%2072188&from_order=4&from_area=aarea02&from_idex=2
[16] 高血壓解惑(Q&A)
https://www.cth.org.tw/?aid=408&pid=2&page_name=detail&iid=873
[17] 高血糖會破壞動脈血管 易引發糖尿病患器官病變
https://www.commonhealth.com.tw/blog/2932
[18] 代謝症候群 - 衛生福利部國民健康署
https://www.hpa.gov.tw/pages/list.aspx?nodeid=221
[19] 代謝症候群增長率45歲後中年女性是男性的2倍 - 自由健康網
https://health.ltn.com.tw/article/breakingnews/3249351
[20] 黃萱1 林毅欣2 1馬偕醫院 外科部 2臺安醫院 內科部內分泌暨新陳代謝科高齡
(65 歲以上) 第 2 型糖尿病患者之處置:文獻回顧DOI:10.6314/JIMT.201812_29(6).04
[21] 糖尿病檢查方法有哪些?快來看看醫師怎麼說?
https://www.cxl.tw/the-test-of-diabet/
[22] 篩檢方法- 糖尿病 - 基隆市e化健康博物館
https://chmuseum.klchb.klcg.gov.tw/web/Content/Content.aspx?c0=167
[23] Kushwaha S, Srivastava R, Jain R, Sagar V, Aggarwal AK, Bhadada SK, Khanna P. Harnessing machine learning models for non-invasive pre-diabetes screening in children and adolescents. Comput Methods Programs Biomed. 2022 Nov;226:107180. doi: 10.1016/j.cmpb.2022.107180. Epub 2022 Oct
[24] Dinh A, Miertschin S, Young A, Mohanty SD. A data-driven approach to predicting diabetes and cardiovascular disease with machine learning. BMC Med Inform Decis Mak. 2019 Nov 6;19(1):211. doi: 10.1186/s12911-019-0918-5. PMID: 31694707; PMCID: PMC6836338.
[25] Makroum MA, Adda M, Bouzouane A, Ibrahim H. Machine Learning and Smart Devices for Diabetes Management: Systematic Review. Sensors (Basel). 2022 Feb 25;22(5):1843. doi: 10.3390/s22051843. PMID: 35270989; PMCID: PMC8915068.
[26] Ellahham S. Artificial Intelligence: The Future for Diabetes Care. Am J Med. 2020
Aug;133(8):895-900. doi: 10.1016/j.amjmed.2020.03.033. Epub 2020 Apr 20. PMID: 32325045.
[27] 糖尿病 - 衛生福利部國民健康署
https://www.hpa.gov.tw/Pages/List.aspx?nodeid=359
[28] 康健雜誌 特刊的26期(2012).Retrieved from
https://www.commonhealth.com.tw/article/72096
[29] 血管硬化的檢查與治療 - 書田泌尿科眼科診所
http://www.shutien.org.tw/dr_doc_detail.aspx?bookid=1104
[30] 糖尿病患傷口難癒合 併發症多與微循環障礙有關
https://healthmedia.com.tw/blog/personal_news_detail.php?id=2931
[31] 腹部肥胖恐增心血管風險醫:這1時間點的女性最危險!
https://www.commonhealth.com.tw/article/83842
[32] 什麼是代謝症候群?哪些人容易發生?醫師詳解代謝症候群症狀、如何避免及緩解
https://health.udn.com/health/story/5974/6364789
[33] 代謝症候群與心血管疾病 - 臺灣醫學會
http://www.fma.org.tw/2011/S-1-2.html
[34] 糖尿病患傷口難癒合併發症多與微循環障礙有關 - NOW健康
https://healthmedia.com.tw/blog/personal_news_detail.php?id=2931
[35] 傷口照護 - 芊城實業
https://www.qcm.com.tw/%E5%82%B7%E5%8F%A3%E7%85%A7%E8%AD%B7
[36] 成人(20歲以上)代謝症候群之判定標準(2007 台灣)
https://www.hpa.gov.tw/Pages/Detail.aspx?nodeid=639&pid=1219
[37] 什麼是代謝症候群?專科醫師帶你認識代謝症候群的定義以及5大指標
https://www.healthsp.org/article/metabolic_syndrome
[38] 代謝症候群對國人的影響- showmt - UNREGISTERED VERSION
https://www.scmh.org.tw/dept/lab/show/Preview/page-140.html
[39] 邱昱植(2022)。結合穿戴式脈波量測、機器學習與脈波分布區間分析(PDA)應用於肌少症 早期評估及探討。國立臺灣科技大學醫學工程研究所碩士論文,台北市。
[40] 楊長仁(2020)。穿戴式脈波量測系統結合機器學習於社區民眾血管狀態評估。國立臺灣科技大學醫學工程研究所碩士論文,台北市。
[41] 翁婉玲(2021)。穿戴式脈波量測系統結合機器學習於失智症及社區民眾心智健康狀態評估。國立臺灣科技大學醫學工程研究所碩士論文,台北市。
[42] 陳均曄(2019)。非侵入式循環訊號量測系統應用於不同癌症之偵測與評估。國立臺灣科技大學醫學工程研究所碩士論文,台北市。
[43] 楊舒涵(2015)。都卜勒微循環血流與動脈血壓波指標 應用於代謝症候群之偵測與評估。國立臺灣科技大學醫學工程研究所碩士論文,台北市。
HH
Wu LW, OuYoung T, Chiu YC, Hsieh HF, Hsiu H. Discrimination between possible sarcopenia and metabolic syndrome using the arterial pulse spectrum and machine-learning analysis. Sci Rep. 2022; 12(1): 21452.
Hsiu H, Lin SK, Weng WL, Hung CM, Chang CK, Lee CC, Chen CT. Discrimination of the Cognitive Function of Community Subjects Using the Arterial Pulse Spectrum and Machine-Learning Analysis. Sensors (Basel). 2022; 22(3): 806.
Hsiu H, Liu JC, Yang CJ, Chen HS, Wu MS, Hao WR, Lee KY, Hu CJ, Wang YH, Fang YA. Discrimination of vascular aging using the arterial pulse spectrum and machine-learning analysis. Microvasc Res. 2021; 139: 104240 (SCI).
Lin SK, Hsiu H, Chen HS, Yang CJ. Classification of patients with Alzheimer's disease using the arterial pulse spectrum and a multilayer-perceptron analysis. Sci Rep. 2021;11(1):8882.
Lin FC, Hsiu H, Chiu HS, Chen CT, Hsu CH. Characteristics of pulse-waveform and laser-Doppler indices in frozen-shoulder patients. Biomed Signal Process Control. 2020; 56: 101718.
Lin Y-J, Lee C-C, Huang T-W, Hsu W-C, Wu L-W, Lin C-C, Hsiu H*. Using Arterial Pulse and Laser Doppler Analyses to Discriminate between the Cardiovascular Effects of Different Running Levels. Sensors. 2023; 23(8):3855. https://doi.org/10.3390/s23083855 (SCI)
Chen CC, Chang CK, Chiu CC, Yang TY, Hao WR, Lin CH, Fang YA, Jian W, Hsu MH, Yang TL, Liu JC*, Hsiu H*. Machine Learning Analyses Revealed Distinct Arterial Pulse Variability According to Side Effects of Pfizer-BioNTech COVID-19 Vaccine (BNT162b2). J Clin Med. 2022 Oct 17; 11(20): 6119. doi: 10.3390/jcm11206119. PMID: 36294440; PMCID: PMC9605457. (SCI)
Wu Z, Yu S, Zhang H, Guo Z, Zheng Y, Xu Z, Li Z, Liu X, Li X, Chen S, Zhang J, Tao L, Guo X. Combined evaluation of arterial stiffness, glycemic control and hypertension for macrovascular complications in type 2 diabetes. Cardiovasc Diabetol. 2022 Nov 28;21(1):262.
Kimoto E, Shoji T, Shinohara K, Inaba M, Okuno Y, Miki T, Koyama H, Emoto M, Nishizawa Y. Preferential stiffening of central over peripheral arteries in type 2 diabetes. Diabetes. 2003 Feb;52(2) 448-452.
Bonarjee VVS. Arterial Stiffness: A Prognostic Marker in Coronary Heart Disease. Available Methods and Clinical Application. Front Cardiovasc Med. 2018 Jun 11;5:64.
McKeage K, Goa KL. Insulin glargine: a review of its therapeutic use as a long-acting agent for the management of type 1 and 2 diabetes mellitus. Drugs. 2001;61(11):1599-624.
Zhang X, Ye R, Yu C, Liu T, Chen X. Correlation Between Non-insulin-Based Insulin Resistance Indices and Increased Arterial Stiffness Measured by the Cardio-Ankle Vascular Index in Non-hypertensive Chinese Subjects: A Cross-Sectional Study. Front Cardiovasc Med. 2022 Jul 5;9:903307