Classification of Sleep Stage for Human Wellness from Single-Channel EEG Using Convolutional Neural Network of Deep Learning
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
In this paper, a model is proposed for automated sleep stage scoring, using Convolutional Neural Network (CNN) and transfer learning by the establishment of raw single-channel electroencephalogram (EEG). In most of the experiments in literature, hand-engineered work has been done based on features that require earlier sleep study analysis. Only a few studies encode time-related information, like rules for the transition. It is an important task in recognizing the next stages of sleep into the drawn-out features. In the proposed model, CNN is used to draw out time-independent features. In addition, to learn the rules of automatic transition between the EEG epochs of various sleep stages, Long-Short Term Memory denoted (LSTM) is used. The implemented training algorithm comprises two steps that helped the proposed model to train effectively. Different single-channel EEGs like Pz-Oz and Fpz-Cz, are used in the phases of training and testing of the proposed model. These channels are available from an openly available dataset that consists of data of 50 healthy persons with totally distinct characteristics such as scoring standards (American Academy of Sleep Medicine (AASM) and Rechtschaffen and Kales (R&K)), with sampling rates. The proposed model for the Fpz-Cz channel achieved a high mean F1 score (86%, range 80–93%), mean accuracy (87%, range 84–89%), and overall accuracy (93.9%, range 90–97%) over every subject. In comparison with state-of-the-art methods, it can be seen that the results of the proposed model are better and the execution is stable throughout classes.
Fields of Study & Subjects
EEG and Brain-Computer Interfaces
Emotion and Mood Recognition
ECG Monitoring and Analysis
Publication details
- Journal / source
- CRC Press eBooks
- Pages
- 246–262
- Publication date
- 31 March 2026
- Language
- EN
- Document type
- Book chapter