基于心率变异分析的睡眠分期方法

The Sleep Staging Based on HRV Analysis

利用隐马尔可夫模型(HMM)对心率变异性(HRV)进行分析,识别HRV在不同睡眠分期的模式变化,从而推算出相应的睡眠分期。在信号处理的过程中采取了一定的措施降低个体差异对分析的影响;在特征提取中还考虑了HRV中超低频分量和睡眠的关系。由于心率信号的提取对睡眠几乎没有任何干扰,因此,本文提出的睡眠分期方法可以较好地反映受试者在自然条件下真实的睡眠状况,实验证明,该方法简单可行,其睡眠分期的结果和人工分期相比的符合率可以满足很多睡眠监测场合的需要,尤其适用于健康人常年的睡眠监测。

In order to deduce the sleep stages from heart rate, we analyze the heart rate variability （HRV） with hidden Markov model （HMM） for the identification of different characters of HRV within different sleep stages. Special technique is used to compensate the individual diversity. The relationship between the sleep stage and the ultra-low frequency components of HRV is also considered. Since the detection of heart rate hardly disturbs the sleep, the proposed method provides a simple approach to evaluating the sleep stage without disturbing the sleep. Our experiments have proved that this method meets the requirements of wide applications, especially the requirement of routine use in monitoring the normal subjects＇ sleep.