Feature Extraction and Classification for Short-Time Sleep

The physiological data recorded during the day time short sleep were analyzed. The objective was to extract the characteristics of sleep stages and realize the automatic determination of sleep stages. Firstly, electroencephalography (EEG) and other physiological data of 20~30 min day time were acquired synchronously. Secondly, fast Fourier transform (FFT) was utilized for the spectral analysis and feature extraction. Finally, support vector machine (SVM) was adopted for automatic determination for short-time sleep data. It was shown from the experimental results that FFT with SVM can achieve better results in the study of short-time sleep stages.Hence, the obtained feature extraction and classification results can be utilized as the assistant information for day time short sleep assessment.