| Summary: | 碩士 === 國立臺北科技大學 === 電子工程系 === 106 === This paper presents a novel earphone-type physiology signals monitor system. Proposed that an optical bio-signal sensor be placed on the anti-tragus region of the human ear to obtain a photoplethysmography (PPG) signal for calculating the instantaneous heart rate and blood oxygen saturation. An infrared thermopile sensor be placed on the ear canal to obtain body temperature. Finally, a real-time monitoring platform is implemented on the smart device to display and record the users physiological parameters.
Compared with the current wrist-type monitoring device nature of the ear is cartilage and microvessels not have too much motion artifact during human activity. Sensors placed in the ear are also less susceptible to environmental light sources or ambient temperature, so environmental noise that must be filtered is also significantly reduced. Experiments of comparing the stability of the wrist and ear the T-test analysis showed that the ears were significantly better than the wrists.
Use an acceleration sensor to decision which algorithm will be used. Static time domain algorithm is used when the human body is at rest. To avoid noise interference using dynamic adaptive time window to analysis physiological parameter. Dynamic frequency domain algorithm with STFT is used when the human body is at action. The algorithm architecture ensures the system performance and the goal of accurate calculation of physiological parameters.
A total of ten male subjects were subjected to experimental reception, and the heart rate, blood oxygen concentration and body temperature were measured in a static sitting position for ten minutes. The medical grade Biopac MP150 was used as the verification standard. The static experiment results showed that the heart rate averaged Pearson correlation coefficient r is greater than the strong correlation of 0.85, the average error is -1.227±1.558 (BPM).The mean error of blood oxygen concentration is -0.221±1.045 (%). The body temperature is -0.148±0.375 (°C). The dynamic verification consisted of walking and running for ten minutes, and using the portable ECG recorder as the verification device. The average Pearson correlation coefficient r of the dynamic heart rate experiment was greater than 0.95, and the average error was -1.01±6.406 (BPM).
|
|---|
