Using the Support Vector Machine for Heart Rate Signal Analysis To Quantify the Sleep Stages

• Main Authors: CHU, PING-HUNG, 朱秉宏
• Other Authors: LIN, YUE-DER
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/ar94hr

碩士 === 逢甲大學 === 生醫資訊暨生醫工程碩士學位學程 === 106 === With the development of society, lack of sleep and sleep disorders have become common problems for Chinese people. To prevent sleep or treat sleep problems. It must be able to track sleep quality. In the past, to achieve this goal, you need to rely on sleep labs. Put on a brain wave cap and multiple instruments . Not only expensive, but there is also a lot of pressure on the measurement experience. This study switched to a heart rate signal that is easy to measure. And get heart rate characteristics through HRV (heart rate variability) . Then determine the sleep stages via the SVM (support vector machine) classifier to quantify sleep quality. This study used the heart rate signal of the MIT-BIH polysomnographic database. The database records more than 80 hours of brain waves, ECG (electrocardiography)and other signals under the standard monitoring system. And the sleep stage is marked by the clinician every 30 seconds during a continuous and stable sleep phase. This study take 30-second and 5-minute heart rate signal. Through HRV analysis to calculate a variety of time domain, frequency domain, and nonlinear heart rate characteristics. Using factor analysis and t-test to filter features. Finally, the feature values are thrown into the SVM for training. Try to improve the correcting rate of SVM judgment. At last successfully use of heart rate signals instead of brain waves to quantify sleep quality. Keywords: sleep quality, sleep stages, SVM (support vector machine), HRV (heart rate variability)