模仿專家行為模式的睡眠分析系統

Abstract

本研究介紹了睡眠分析的另一個可能方向。藉由收集受測者的三組腦波訊號、兩組眼動訊號以及肌電訊號進行特徵波型的分析與睡眠分期決策。這個演算法模擬臨床技師以肉眼辨識特徵腦波，由時域分析腦波訊號以求更貼近專家的辨識結果；分期策略也不是單單只看某特徵的出現與否就能做出決策，除了依標準睡眠腦波的判斷規則外，對於部份沒有特徵出現的頁面，還會參照前後片段決定當下的片段應如何分期。在本論文的最後呈現了四位受測者的睡眠分期之分析結果，並與國外幾篇重要的研究做了分析結果的比較。

A novel algorithm for determining subject’s sleep stages based on signals collected from three EEG channels, two EOG channels and one EMG channels were presented. The algorithm mimics the iterative hypothesis-testing approach often used by clinicians, analyses EEG signal from the time domain in order to identify more closely the results of the experts. Sleep staging strategy is not just look at the presence or absence of a feature, however, followed the standard judge rules of sleep staging and for some pages which appears is not obviously, algorithm determines the current stage by relationship between before and after epoch. This study presented the staging results of four subjects and compared the results with other important studies.