「世大智科/天才家居」-我們創業囉
Contact Professor: Yeh-Liang Hsu (徐業良)

九十三學年度元智大學機械工程研究所阮致維碩士論文

Master thesis by C. W. Reng, Mechanical Engineering Department, Yuan Ze University, 2005

93碩士論文:以外顯行為訊號判讀睡眠階段

  根據「亞洲睡眠協會」近年來的睡眠品質調查報告顯示台灣成年人約有一半的比例睡不好,其中四十五歲以上族群,有高達八成自認為存在睡眠問題 。 睡眠品質評估工具分成三類,但一般來說不適合大眾在居家環境下以客觀數據自我評估睡眠品質。
  開發在居家環境中可長期持續地評估睡眠品質工具,是我們的目標。 「可攜式居家睡眠品質評估系統」之研究,能在不干擾使用者睡眠的前提下,在居家睡眠環境中長期持續性地監測呼吸頻率、身體活動、鼾聲等三項外顯行為,分析受測者睡眠歷程,睡眠品質四項重要指標。其中評估系統中,以判讀睡眠階段之演算法最為重要,本研究即嘗試開發此評估系統中之演算法。本研究蒐集了14位受測者以PSG蒐集之整晚睡眠資料,加以分析、處理,建立外顯行為與睡眠階段之關係。
  接著本研究即嘗試發展一套判讀睡眠階段演算法,以呼吸頻率變異係數和腿部動作頻率兩項外顯行為訊號判讀睡眠階段,能將整晚睡眠區分為清醒、深眠、淺眠、REM等四個階段。最後演算法判讀結果經與PSG判讀結果比對,判讀符合率為56.44%。

Using Signals from Explicit Body Activities to Differentiate Sleep Stages

  Investigations show that more than 50% of adults in Taiwan do not sleep well. Especially for the age groups over 45, many people feel that they have sleep disorders. However, current tools of sleep quality assessment are not suitable for people to do long – term sleep quality assessment in the home environment.
  This research is part of the project for developing a sleep quality assessment system in a home environment. This system uses signals from explicit body activities behaviors such as respiration frequency, body activities and snores, to different whole-night-sleep into 4 sleep stages, awake, light sleep, deep sleep, and REM. The critical part of this system is the algorithm, which differentiates sleep stages using signals from explicit body activities to three body’s behaviors.
  This research first obtained sleep data from 14 testers with Polysomnography (PSG). From this data, the relations between explicit body activities and sleep stages are analyzed. An algorithm is developed to differentiate sleep stages with respiration frequency and frequency of leg movements to differentiate sleep into four sleep stages including awake, deep sleep, light sleep and REM. The results of algorithm were compared to the results of PSG, and the success rate is 56.44%.

Academic Thesis
Presentation Files
Last Updated:2010/6/8