Scholarlink

<p>Nearly one-third of a person's lifespan is spent sleeping. The quality of sleep has a direct impact on a person's ability to function mentally and physically. Poor sleep habitually degrades the performance of individuals; therefore, adequate night sleep is essential. Sleep stage classification (SSC) plays a vital role in sleep disorder analysis to diagnose an individual's psychological and physical health. This paper offers SSC based on Deep Convolution Neural Network (DCNN) using multichannel polysomnograms (PSGs) such as electrooculogram (EOG), electroencephalogram (EEG), electromyogram (EMG) signals to improve feature distinctiveness and connectivity of sleep stage. Further, the Generated Adversarial Neural Network (GAN) is used for data augmentation to reduce the data scarcity problem. Additionally, an Analytical Hierarchical Process (AHP) is applied to choose the optimal hyper-parameters of the proposed DCNN using various performance metrics as the quality metrics. The results of the proposed SSC are evaluated on the SleepEDF dataset using accuracy, precision, recall, and F1-score. The comparative analysis states that the proposed scheme provides noteworthy improvement over SSC's traditional state of arts.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multichannel sleep stage analysis using deep convolution neural network and analytical hierarchical process

  • 期刊论文

  • 发表日期：

    2025年4月14日

查看全文
  • Anjali W. Pise,
  • Priti P. Rege

摘要

Nearly one-third of a person's lifespan is spent sleeping. The quality of sleep has a direct impact on a person's ability to function mentally and physically. Poor sleep habitually degrades the performance of individuals; therefore, adequate night sleep is essential. Sleep stage classification (SSC) plays a vital role in sleep disorder analysis to diagnose an individual's psychological and physical health. This paper offers SSC based on Deep Convolution Neural Network (DCNN) using multichannel polysomnograms (PSGs) such as electrooculogram (EOG), electroencephalogram (EEG), electromyogram (EMG) signals to improve feature distinctiveness and connectivity of sleep stage. Further, the Generated Adversarial Neural Network (GAN) is used for data augmentation to reduce the data scarcity problem. Additionally, an Analytical Hierarchical Process (AHP) is applied to choose the optimal hyper-parameters of the proposed DCNN using various performance metrics as the quality metrics. The results of the proposed SSC are evaluated on the SleepEDF dataset using accuracy, precision, recall, and F1-score. The comparative analysis states that the proposed scheme provides noteworthy improvement over SSC's traditional state of arts.