工學院: 應用力學研究所指導教授: 劉佩玲盧岳璟Lu, Yueh-ChingYueh-ChingLu2017-03-062018-06-292017-03-062018-06-292016http://ntur.lib.ntu.edu.tw//handle/246246/277085本研究旨在期望利用一種簡單的方式判斷使用者的睡眠穩定度，故本研究嘗試利用穿戴式裝置常之光學式血管體積變化訊號(PPG)來做為判斷睡眠穩定度的生理特徵，並且結合心肺耦合分析法(CRC)來演算分析並與睡眠多項生理檢查儀(PSG)中的循環交替圖譜(CAP)這種被研究出與睡眠穩定度相關性極高的腦波圖譜進行相關性分析，期待能找出與睡眠穩定程度的關聯性。 我們利用PSG中計算血氧濃度(S_p O_2)的PPG訊號進行研究，並僅利用PPG訊號截取出心跳變異率時間序列(NNI)與PPG推導之呼吸特徵訊號(PDR)進行心肺耦合分析，因此不需要穿戴呼吸感測器。 接著將CAP的腦波判讀階段依照睡眠穩定程度分為三個階段，非循環交替圖譜(NCAP)、CAP與清醒與快速眼動期(WREM)。為了與上述CAP階段做比較，我們利用Welch方法計算NNI及PDR之功率頻譜密度(PSD)並相互耦合，其中高頻耦合能量(EHF, f =0.1~0.4 Hz)、低頻耦合能量(ELF, f =0.01~0.1 Hz)與ELF /EHF比例(Lo/Hi ratio)等，便是用來與CAP階段比較的相關特徵。 本研究共21位受試者，其中非睡眠呼吸中止症5位、輕中度患者6位與重度患者10位。其結果呈現非OSA患者準確率不佳(ACC_(PPG,NOSA)=38.37 %)，而OSA患者整體的準確度也不好(ACC_(PPG,OSA)=45.31 %)，但在判讀睡眠相對穩定階段即NCAP階段的準確率表現相對好很多(ACC_(PPG,NCAP,OSA)=75.15 %)。 由於NCAP判別的準確率較高，本研究在OSA患者分組利用一個穩定睡眠NCAP發生時間除以總睡眠時間(NCAP/TST)即NCAP Ratio這個能代表穩定睡眠在整夜中比例的指標來與睡眠穩定指標CAP發生率(CAP Rate)做比較，並利用迴歸分析發現此方法與人工判讀之CAP Rate呈現負相關的趨勢，代表此PPG方法與NCAP Ratio的指標在判斷睡眠穩定度上在未來是有機會可行的且同時也具有一定程度的生理意義。 比較先前的研究方法，本研究可以利用一個更簡單量測的PPG訊號來達成與心電圖訊號(ECG)相同的睡眠穩定度判讀趨勢，未來若是要應用在居家量測，我們認為PPG訊號將會是一個更好的選擇。The main purpose of this study is to use an easy method to analyze users sleep stability. Nowadays, lots of wearable devices have already used photo-plethysmography signal (PPG) to measure physiological characteristics of sleep. So we choose PPG signal to evaluate sleep stability. And we compare the Cardiorespiratory Coupling (CRC) result with cyclic alternating pattern (CAP) in order to find out the correlation of sleep stability. In our study, the PPG data recorded from Polysomnography (PSG). And our study only uses PPG signal to detect the heartbeat interval (NNI) and PPG derived respiratory signal (PDR). So we do not need to wear airflow sensor to measure the characteristics of respiration. The CAP stages in our study are divided into 3 types (NCAP, CAP, and WREM) according to the level of sleep stability. Then we use Welch’s method to calculate the power spectral density (PSD) of NNI and PDR. When we coupling them together, we’ll get high frequency band power energy (EHF, f =0.1~0.4 Hz), Low frequency band power energy (ELF, f =0.01~0.1 Hz), and ELF/ EHF (Lo/Hi ratio). The foregoing are the characteristic we compare with CAP stages. We have 21 subjects which were divided into 3 groups: 5 normal subjects, 6 mild and moderate subjects, and 10 severe subjects. And the results show that the normal subjects do not have an obvious characteristic to compare with CAP (ACC_(PPG,NOSA)=38.37 %). Next, we analyze the sleep apnea subjects. The total accuracy of sleep apnea subjects is not a good result (ACC_(PPG,OSA)=45.31 %). But we find the detection of stage non-cyclic alternating pattern (NCAP) have a nice performance (ACC_(PPG,NCAP,OSA)=75.15 %). In order to use the detection of NCAP to determine sleep stability. We use NCAP time divided by total sleep time (NCAP Ratio). Finally, we find the result of NCAP Ratio shows a negative correlation with CAP Rate. So we think this method might have the potential to evaluate sleep stability. Compared to the previous research, the method of PPG’s measurement is easier than ECG signal. So we think this PPG method might be a better choice for Home Healthcare in the future.論文使用權限: 不同意授權睡眠呼吸中止症睡眠穩定度心肺耦合分析光學式血管體積變化訊號循環交替圖譜Sleep Apnea SyndromeSleep StabilityCardiorespiratory Coupling AnalysisPhoto-PlethysmographyCyclic Alternating Patternthesis10.6342/NTU201602369