基于双向递归神经网络的单通道脑电图睡眠分期研究

摘要

Sleep staging has important research and application value in the field of neuroscience and psychiatry.The traditional artificial sleep staging method is inefficient and easily influenced by subjective factors.Recently,with the development of machine learning method,automatic sleep staging method has obtained certain achievements,but there are still some shortcomings.In order to achieve effective automatic sleep stages,a sleep staging method based on the bidirectional recursive neural network is proposed for single channel EEG.It gives full play to the bidirectional recursive neural network temporal expression ability and characteristics of excellent learning ability.Through experiment in the case of single channel EEG,sleep staging accuracy can reach 95 ％.Results show that this method can effectively improve the accuracy of automatic sleeping stages and has a good application prospect.%睡眠分期在神经学和精神学等领域具有重要的研究与应用价值.传统的人工睡眠分期方法效率低下并且易受主观因素影响.近年来,随着机器学习方法的发展,自动睡眠分期方法取得了一定的成果,但还是存在诸多不足.为了有效地实现自动睡眠分期,提出了基于双向递归神经网络的单通道脑电图睡眠分期方法.该方法充分发挥了双向递归神经网络优异的时序表达能力和特征学习能力.经过实验得出,在单通道脑电图的情况下,睡眠分期准确率可达到95％.因此,该方法能够有效地提高自动睡眠分期的准确率,具有良好的应用前景.