睡眠分期的符号转移熵分析

目的睡眠分期是衡量睡眠质量和诊治睡眠障碍性疾病的重要途径，转移熵是一个量化2个序列相关程度的参数。本文将基于符号化技术的符号转移熵首次应用在睡眠分期研究中，克服了以往方法对参数之间协调性要求高以及对噪声敏感的缺点。方法通过提取相同个体相同时刻的清醒期和非快速眼动睡眠I期的EEG、ECG信号，分别进行符号化、相空间重构后，计算符号转移熵，对两个睡眠阶段的符号转移熵进行t检验及多样本验证。结果实验结果表明清醒期符号转移熵大于非快速眼动睡眠I期的符号转移熵。经t检验表明这两个阶段的符号转移熵值有显著性差异，并通过多样本验证。随着睡眠加深，身体单元不断偶合，符号转移熵减小，与理论分析相符合。结论清醒期和非快速眼动睡眠I期的符号转移熵很好地体现了睡眠状态的变化，因此符号转移熵可用于睡眠分期，并成为研究睡眠自动化分期的极具潜力的分析工具。

Symbolic transfer entropy analysis of sleep stage classification

Objective Sleep stage classification is important to the eva｜uation of sleep quality, the diagnosis and treatment of sleep disorders. Transfer entropy is a parameter to measure the relevance of two time sequences symbolic transfer entropy（STE）, which is based on the technique of symbolization, may be the first application in the study of sleep stage classification. This method overcomes the drawbacks of requirement for the coordination between parameters and the sensitivity to noise contributions. Methods The EEG and ECG signals about the wake stage and the first stage of non-rapid eye movement sleep are extracted from the same people at the same time. After symbolic and space reconstruction, we compute the STE and test by t test with multi-samples. Results The STE of wake stage is larger than that of the first stage of non-rapid eye movement sleep and the difference between the two sleep stages is significant in t test. Brain ceils and heart ceils continue to be coupling, the STE become smaller as the sleep stage is deepening. Therefore, the experiment results are consistent with the theory. Conclusions The STE about the wake stage and the first stage of non-rapid eye movement sleep reflects the changes of sleep stages and can be applied in sleep stage classification. STE may be a promising tool for the analysis of automatic sleep classification.