文献类型：期刊文献

中文题名：结合状态转移规则的深度睡眠分期模型

英文题名：Deep automatic sleep staging model integrated with state transition rules

作者：马家睿[1];王蓓[1];金晶[1];王行愚[1]

机构：[1]华东理工大学化工过程先进控制和优化技术教育部重点实验室,上海200237

年份：2020

卷号：41

期号：10

起止页码：2878

中文期刊名：计算机工程与设计

外文期刊名：Computer Engineering and Design

收录：CSTPCD;;北大核心:【北大核心2017】；

基金：国家自然科学基金项目(61773164、91420302);上海市自然科学基金项目(16ZR1407500)。

语种：中文

中文关键词：自动提取特征;深度神经网络;睡眠分期;状态转移;Dijkstra算法

外文关键词：automatic feature extraction;deep neural network;sleep staging;state transition;Dijkstra algorithm

摘要：为解决传统睡眠分期特征需手工设计且忽略睡眠状态变换的前后关联性和规律性的问题,设计一种结合状态转移规则的深度睡眠分期模型。通过添加残差网络加深卷积神经网络层数,自动提取信号的高维特征,对睡眠状态进行分类,结合睡眠状态变换规律,设计状态转移规则纠正分类结果。实验结果表明,该模型有效可行,为睡眠相关疾病的诊断和治疗提供了可行的辅助判读方法。
A deep neural network integrated with sleep state transition rules was developed to solve the shortcomings of the ma-nual feature extraction and the problem of less attention on the regular sleep state transformation.The residual network was designed to deepen the network layer and extract the high-level features for sleep stage classification.State transition rules were constructed based on the regular dynamic transition among the sleep stages to correct the classification results.The obtained results show that the model is effective and feasible.It can be a useful and realizable assistant sleep stage interpretation tool for the diagnosis and treatments of sleep related disorders.