Reflective photoplethysmography (PPG) signals are widely applied on many wearable devices to detect physiological signs, such as heart rate, blood pressure and peripheral oxygen saturation (SpO2). Current research shows that the contact force between sensors and skin can effectively influence PPG signals. This work mainly focuses on identifying the major aspects affected by contact forces. Based on an open-source database, WF-PPG, the signal qualities under different contact force were compared to evaluate the influence on high-frequency noise, baseline wandering, and motion artifacts. Additionally, the datasets were used to train a linear regression model and a support vector regression (SVR) model for SpO2 prediction. The results of signal quality analysis show that as contact force increases the reflective PPG will have increased high-frequency noise and be more affected by motion artifacts. Moreover, the parameters of trained linear model indicate that the optimal gain and offset will change with variations in contact force. When contact force is include, the SVR model outperforms the model without contact force. In conclusion, the contact force can affect the waveform of reflective PPG and further influence the measurement of SpO2. Thus, detecting contact force simultaneously can actually improve the accuracy of SpO2 measurement.

reflective PPG,contact force,SpO2