| Abstract [eng] |
This thesis investigates an ear-worn human sleep posture and heart rate variability monitoring system based on the photoplethysmography (PPG) method. The relevance of the study is driven by the increasing need for comfortable, non-invasive, and everyday-use physiological monitoring systems capable of operating during sleep. Traditional systems, such as electrocardiography and polysomnography, often require complex equipment and limit user mobility and comfort. During the study, a scientific literature review was performed, and the experimental “OpenEarable 2.0” platform together with the MAXM86161 PPG sensor was selected. Experimental measurements were performed using four participants, each repeating the measurement protocol twice, with recordings acquired from both the left and right ears. In total, eight experimental recordings were collected, consisting of 112 analysis segments and 336 PPG signal segments from infrared (IR), red, and green optical channels. Photoplethysmographic signals were recorded in different body positions: standing, lying on the back, lying on the left side, and lying on the right side. A signal processing algorithm was developed in the MATLAB environment for PPG signal quality assessment, sleep posture identification, and heart rate variability analysis. To evaluate signal quality, a Signal Quality Index (SKI) was developed, consisting of signal-to-noise ratio, perfusion index, heartbeat regularity, pulse morphology, template matching, and entropy components. Additionally, the “NeuroKit2” library was integrated into the MATLAB environment for heart rate variability parameter calculation. The results of the study showed that the infrared (IR) channel provided the highest PPG signal quality and the greatest stability across different body positions. The highest average SKI value was observed in the IR channel while lying on the back (SKI ≈ 0.71), whereas the lowest value was recorded in the green channel during standing posture (SKI ≈ 0.50). Signals with SKI values higher than 0.7 were considered sufficiently reliable for further pulse and heart rate variability analysis. It was determined that body posture affects both signal quality and heart rate variability parameters. The highest signal stability was observed while lying on the back, whereas the greatest amount of motion artifacts occurred during side-lying positions. RMSSD and SDNN parameter analysis demonstrated changes in autonomic nervous system activity depending on body posture. Additionally, the influence of sensor placement on PPG signal quality was evaluated by comparing signals recorded from the left and right ears. No significant differences between ears were observed; however, when lying on the side, the compressed ear exhibited increased signal noise and lower SKI values. The obtained results demonstrated that an ear-worn PPG system can be applied for human sleep posture and heart rate variability monitoring and has potential for the further development and practical application of non-invasive sleep monitoring technologies. |
