论文部分内容阅读
睡眠分期是研究睡眠及相关疾病的基础,是完成睡眠质量评估的前提。为实现有效睡眠自动分期,本文提出将能量特征和最小二乘支持向量机(LS-SVM)相结合的方法。先利用FIR带通滤波器提取Pz-Oz导睡眠脑电信号的特征波,获得能量特征,并与小波包变换方法相比较;然后用LS-SVM分类器进行模式识别,最终实现睡眠自动分期。实验表明,本文所提出的基于能量特征和LS-SVM的自动睡眠分期方法简单、有效,平均正确率达88.89%,具有很好的应用前景。
Sleep stage is to study the basis of sleep and related diseases, is to complete the premise of the quality of sleep assessment. In order to realize the automatic staging of effective sleep, a method of combining energy features with least square support vector machine (LS-SVM) is proposed in this paper. Firstly, the characteristic wave of Pz-Oz-induced sleep-induced brain electrical signal was extracted by FIR band-pass filter, and the energy characteristics were obtained and compared with the wavelet packet transform method. Then LS-SVM classifier was used for pattern recognition and eventually sleep automatic staging. Experiments show that the proposed method based on energy features and LS-SVM automatic sleep staging method is simple and effective, with an average correct rate of 88.89%, which has good application prospects.