Intelligent automatic sleep staging model based on CNN and LSTM.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Science

Background:

• Electroencephalogram (EEG) is crucial for diagnosing and treating sleep disorders.
• Conventional EEG feature extraction methods are inefficient and may miss key information.
• Deep learning offers powerful data analysis capabilities for complex biological signals.

Purpose of the Study:

• To develop an efficient and accurate automatic sleep staging method using multi-channel EEG.
• To improve sleep stage recognition accuracy by employing a deep learning approach.
• To validate the proposed method on a recognized sleep dataset.

Main Methods:

• A novel automatic sleep phase method based on multi-channel EEG and deep learning is proposed.
• A Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model is utilized for sleep stage classification.
• Data expansion (DA) is applied for unbalanced data without preprocessing or feature extraction.

Main Results:

• The CNN-LSTM model effectively monitors EEG and electrooculogram (EOG) samples during sleep.
• The proposed method demonstrates high accuracy in sleep stage recognition.
• Experimental results on the MIT-BIH dataset validate the model's effectiveness.

Conclusions:

• The developed EEG-based sleep phase method offers an effective solution for sleep disorder diagnosis and treatment.
• The study highlights the practical application value of deep learning in automatic sleep staging.
• This approach provides a valuable tool for advancing sleep medicine.