Automatic Sleep Quality Assessment for Obstructive Sleep Apnea Patients Based on HRV Spectrum Analysis

Sleep quality has an important role in the health of the human being. There are some risks due to the lack of sleep quality, such as reduced productivity of work, mental, and physical health. Obtrusive sleep apnea (OSA) is a potentially serious sleep disorder. It caused breath repeatedly to stop and start during sleep. Polysomnography (PSG) is the most common approach of sleep quality assessment for OSA patients. However, this method is intrusive and involved a lot of costs. Therefore, this research is proposed to develop an automatic and non-intrusive method of sleep quality assessment for OSA patients. The proposed method used only ECG signal that easy to conduct and record. The HRV spectrum analysis was applied to features extraction and DTB-SVM classifier was used to measure four parameters of sleep quality: Sleep Onset Latency (SOL), Total Sleep Time (TST), Sleep Efficiency (SE), and Delta-Sleep Efficiency (DSE) based on 30-second segments of ECG signals. 25 subjects of OSA patients' database were evaluated in this research. The sleep quality was estimated using automatic sleep stage and then compared with PSG data. Mean absolute errors in SOL, TST, SE, and DSE are 0.6 minutes, 9.6 minutes, 2.26%, 0.85%, respectively. This results implied that the proposed method could be reliably for applied sleep quality assessment in-home and portable system instead of using a multichannel signal for OSA patients.