Application of genetic algorithm on ECG-based features selection for sleep staging

• Main Authors: Yi-heng Wu, 吳宜衡
• Other Authors: none
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/28799871249613553469

碩士 === 國立高雄大學 === 資訊工程學系碩士班 === 103 === 　　Sleep is very important to everyone. However, not everyone can acquire good sleep quality. For the diagnosis, all night polysomnographic (PSG) recordings are usually taken from the patients. The doctor needs to realize the sleep quality and quantity of them. Nevertheless, visual sleep scoring is a time consuming and subjective process. Therefore, developing an automatic sleep scoring method is a very important issue. Due to the disturbance from typical biomedical signals: EEG, EOG, and EMG recording are too huge, the sleep quality scored from those signals is not accurate enough. So our objective of this study is developing an automatic sleep scoring method which only uses the heart rate as the input signal. Although the method using HRV as the input signal is not good enough, the benefits like less disturbance, easy to use and capability of detecting sleep cycle, make it has unlimited potential. 　　We used Genetic Algorithm(GA) to select some suitable features calculated ECG for sleep staging. Combine DHMM which is trained by using the codebook for all testing features. The trained DHMM model is used for sleep staging. Through the evolution of GA and DHMM, better chromosomes or more suitable features are obtained.