Abstract: The method of EEG fuzzy features classification is used to carry out sleep staging. Firstly, EEG preprocessing filters out interference and noise. Secondly, fuzzy entropy algorithm, multiscale entropy algorithm and complexity algorithm are used to extract EEG characteristic parameters. Then the Least Squares Support Vector Machine（LS-SVM） is taken to classify the characteristic parameters, and the sleep process is divided into wakefulness, light sleep, deep sleep, Rapid Eye Movement（REM）, after that sleep staging accuracy is obtained. Finally, 2,000 group sleep EEG samples are tested by the above method, and then staging results of experts are compared with the automatic stage. Taking the complexity as the input of LS-SVM, the average accuracy rate of sleep stage classification is 92.65%, which is higher than the rate of fuzzy entropy or multiscale entropy as the input. The results show that the complexity of the fuzzy feature extraction based on the characteristic parameters can be used as a valid basis for sleep staging. Labor costs is reduced under the premise of ensuring the accuracy.

Key words: sleep staging, fuzzy feature, complexity, multiscale entropy, fuzzy entropy

摘要： 利用脑电信号模糊特征分类的方法对睡眠进行分期研究。首先对脑电信号进行预处理，滤除干扰噪声后使用模糊熵算法、多尺度熵算法以及复杂度算法对脑电信号进行特征参数提取，采用最小二乘支持向量机（the Least Squares Support Vector Machine，LS-SVM）对特征参数进行分类，并将睡眠过程分为清醒期、浅睡期、深睡期和快速眼动期（Rapid Eye Movement，REM），获得分期正确率。最后通过上述方法对2?000组睡眠脑电样本进行睡眠分期测试，与专家人工分期结果进行比对，将复杂度输入到最小二乘支持向量机进行分类的平均正确率是92.65%，高于模糊熵和多尺度熵作为最小二乘向量机的输入时的准确率。基于模糊特征的复杂度提取的特征参数可以作为睡眠分期的有效依据，在保证准确度的前提下，降低人工成本。

关键词: 睡眠分期, 模糊特征, 复杂度, 多尺度熵, 模糊熵