Research On Motion-resistant Heart Rate And Oxygen Saturation Extracting Algorithm Based On Photoplethysmography

Heart rate and oxygen saturation are two very important physiological parameters of human body. Non-invasive and real-time detection of heart rate can not only help people keep their physical condition, but also avoid emergency caused by cardiovascular diseases. In addition, heart rate is one of the most direct indicators of exercise intensity. It can be used to control the amount of exercise to avoid too much pressure on the immune system caused by excessive exercise. Therefore, the real-time and noninvasive detection of heart rate has been paid more and more attention. Oxygen is the material basis of all life activities, and the important substance to for human metabolism. The detection of oxygen saturation in real-time is a key indicator to determine whether the Human respiratory system and circulatory system is normal. The photoplethysmographic(PPG) signal contains a lot of human physiological information, including heart rate and oxygen saturation, so it is widely used to extract the heart rate and oxygen from PPG signal. However, PPG signals are vulnerable to motion artifacts, which interfere strongly with heart rate and oxygen saturation monitoring. In particular, the performance of traditional heart rate and oxygen saturation extracting algorithms is not satisfactory when sudden changes are included in the waveform.To solve the problem, a motion-resistant heart rate extracting algorithm was proposed based on time-varying autoregressive(TVAR) model. Acceleration signal was firstly used to classify PPG signal into two states: PPG in stillniss and PPG in motion. For PPG in stillness, a new method—Mean Intersection Method was proposed. The method has high accuracy and low computational complexity. For PPG in motion, this paper combined a time-varying autoregressive model algorithm, fast Fourier transform(FFT), and acceleration signal. A bandpass filter and an average moving filter were used for signal preprocessing of raw photoplethysmographic signal. Then, the proposed algorithm employed the TVAR modelling approach based on multi-wavelet basis functions to divide the corrupted PPG into different parts. In each part, FFT was applied for spectrum analysis and all the possible heart rates were determined by searching peaks among spectrum. Acceleration signals were used for heart rate optimum selection and heart rate was finally determined. Five healthy people were tested. There were two test modes: resting and motion. Three situations were included in motion modes, namely jumping,waving arm and running. Compared with traditional fast Fourier transform(FFT) method and least mean square(LMS) based heart rate extracting method, the results shows that the proposed method has higher accuracy. In particular, the proposed method is effective for heart rate extracting when sudden changes caused by motion artifacts are included in PPG waveform. The new method has high accuracy and good motion-resistant ability. Combining the new motion-resistant heart rate extracting algorithm and frequency domain approach, we proposed the motion-resistant oxygen saturation extracting algorithm. The result shows that the method has good accuracy and stability.