Scholarlink

<p>This study proposes a deep learning approach using Convolutional Neural Network (CNN) to classify different stages of sleep, provide quality assessment, and understand sleep patterns and disorders. The key features such as frequency, amplitude, and temporal dynamics across multiple time scales are extracted from multichannel electroencephalogram (EEG) data using Adaptive Wavelet Transform (WT). These parameters are adjusted to fit in the range of (−&#xa0;1, 1) to maintain inter-subject consistency, which improves model performance. A Multiple-Input Multiple-Output (MIMO) CNN architecture is employed to effectively process diverse EEG channels and time segments, enabling comprehensive analysis of complex sleep patterns. The CNN model uses the input features to select necessary EEG electrodes and specific time windows of sleep stages that can be compared with the actual labels, providing insights into advanced measures such as sleep efficiency and fragmentation indices. The main contributions consist of more effective feature extraction using Adaptive Wavelet Transform with increased accuracy, enhanced stability, and dynamic normalization. The developed system also equips feedback and outputs for clinical practice, promoting sleep researchers' work and enabling patient sleep monitoring and disease diagnosis at the interface. Additionally, a confusion matrix is used to evaluate the model's performance, providing insights into the accuracy, precision, recall, and F1-score of sleep stage classification.</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

Neuraltapestry: Illuminating Sleep Stages Through EEG Patterns

  • 期刊论文

  • 发表日期：

    2025年3月13日

查看全文
  • S. A. Priyanka,
  • M. Jayanthi,
  • V. Parnika

摘要

This study proposes a deep learning approach using Convolutional Neural Network (CNN) to classify different stages of sleep, provide quality assessment, and understand sleep patterns and disorders. The key features such as frequency, amplitude, and temporal dynamics across multiple time scales are extracted from multichannel electroencephalogram (EEG) data using Adaptive Wavelet Transform (WT). These parameters are adjusted to fit in the range of (− 1, 1) to maintain inter-subject consistency, which improves model performance. A Multiple-Input Multiple-Output (MIMO) CNN architecture is employed to effectively process diverse EEG channels and time segments, enabling comprehensive analysis of complex sleep patterns. The CNN model uses the input features to select necessary EEG electrodes and specific time windows of sleep stages that can be compared with the actual labels, providing insights into advanced measures such as sleep efficiency and fragmentation indices. The main contributions consist of more effective feature extraction using Adaptive Wavelet Transform with increased accuracy, enhanced stability, and dynamic normalization. The developed system also equips feedback and outputs for clinical practice, promoting sleep researchers' work and enabling patient sleep monitoring and disease diagnosis at the interface. Additionally, a confusion matrix is used to evaluate the model's performance, providing insights into the accuracy, precision, recall, and F1-score of sleep stage classification.