Mechanistic Research On The Effect Of Controlled Contact Pressure On The Photoplethsmography Signals

With the development of artificial intelligence and wearable technology,health monitoring devices based on photoplethysmography have become a new research hotspot.Wearable health monitoring devices have received attention from researchers because of their ability to conveniently monitor human vital signs and facilitate the tracking and analysis of trends in human health conditions.However,wearable devices usually use the epidermal tissue at the end of the limb as the detection site,which makes the detection process highly susceptible to sensor contact pressure problems,which in turn affects the accuracy and stability of physiological parameter measurements,and contact pressure has become a bottleneck limiting the application of photoplethysmography.In this study,a controlled contact pressure photoplethysmography acquisition system was designed independently based on photoplethysmography sensor MAX30102 and pressure sensor FSR400.61 volunteers’ photoplethysmography signals were acquired under 6 different fingertip contact pressures in the range of 30mmHg-80mmHg,forming a photoplethysmography data set under different contact pressures.Finally,a CNN-LSTM neural network was applied to construct a blood pressure prediction model to determine the optimal contact pressure range for blood pressure detection by comparing the blood pressure prediction performance of photoplethysmography signals at different contact pressures and to further validate the effect of contact pressure on photoplethysmography morphology.By comparing the photoplethysmography characteristics at different contact pressures,it was found that the perfusion index was significantly higher in the moderate contact pressure range of 50mmHg-60mmHg than at other pressures,including up to 1.78 at 60mmHg.The results of the morphological analysis based on time domain features,amplitude features,ratio features,area features and power features also indicated that the PPG signals at 50mmHg-60mmHg was significantly higher than at other pressures.The morphological features of the PPG signals at 60mmHg pressure are more obvious,such as the amplitude of the systolic wave of the PPG signal,compared to 30mmHg pressure,the peak systolic wave increases by 2.5%at 50mmHg pressure and decreases by 0.6%at 80mmHg pressure,and the signal quality of the PPG signals are higher in the medium pressure range of 50mmHg-60mmHg and more stable and less influenced by pressure factors than the 30mmHg-50mmHg pressure range.The viscoelastic and nonlinear elastic models of the fingertip also explained the effect of contact pressure on arterial vasodilation and contraction.Meanwhile,the performance of the blood pressure prediction model constructed based on CNN-LSTM neural network at different contact pressures also showed more accurate blood pressure prediction results of PPG signals at moderate contact pressures of 50mmHg-60mmHg,with the smallest ME and SDE of diastolic pressure at 50mmHg contact pressure,reaching 0.19mmHg and 2.39mmHg,respectively.Therefore,the research of photoplethysmography morphology under controlled contact pressure improves the measurement accuracy and stability of wearable health monitoring devices,further improves the system of photoplethysmography morphology features under different contact pressures,and it is useful for developing a wearable cuff-less blood pressure measurement system based on photoplethysmography morphology analysis.It has important theoretical value for the development of theoretical system of wearable cuff-less blood pressure measurement based on photoplethysmography morphology analysis.