The Study Of Key Technologies For Non-Invasive Blood Pressure Monitoring Based On Fusion Of Pulse Wave Sensors And Fabric Electrodes

The age of onset of cardiovascular diseases is lowering and the incidence of cardiovascular diseases is rising as a result of societal development,an accelerated pace of work and living,and an intensifying aging of the population.An vital physiological indicator of the human body,blood pressure has a direct correlation to the prevalence of cardiovascular disorders.As a result,monitoring blood pressure is crucial for both preventing and treating cardiovascular illnesses.Likewise,the quick development of neural network technology and intelligent wearable technology makes it possible to continually monitor blood pressure for a considerable amount of time.In this study,a system for non-invasively monitoring blood pressure is developed using a combination of fabric electrode,pulse wave sensor,and neural network technology.The method collects the human photoelectric volume pulse wave(Photoplethysmography,PPG)signal through the sensor,the human v1,v2 double-lead ECG signal through the embroidery structure filled stitch fabric electrode,preprocesses the collected signal,and then employs the neural network model to validate the blood pressure prediction.The following are the paper’s primary contents:(1)To investigate how physiological signals and blood pressure are related.The correlation between blood pressure,the pulse wave signal,and the ECG signal,as well as the selection of the measurement indices to be utilized in the database and model for the experiment.(2)Building and testing a neural network model for monitoring blood pressure.The best performance model is chosen after a variety of models have been built,trained,and compared using RMSE,MAE,and Bland-Altman diagrams.The data from the UCI-BP database is transformed into time series as the training set for the model,which is based on Long Short Term Memory(LSTM)network or bidirectional LSTM network.The RMSE of systolic blood pressure and diastolic blood pressure on the test database set are,4.686 mm Hg 2.694 mm Hg,MAE and 3.688 mm Hg,MAE 1.857 mm Hg.(3)A system for non-invasive blood pressure monitoring was designed and experimentally verified.The hardware design uses an NRF52832 master chip,a MAX30102 sensor,and fabric electrodes with an embroidered structure and stitch filling surface to capture pulse wave signals and electrocardiogram(ECG)signals,respectively.Denoising,data normalization,and data resampling utilizing wavelet transform and principal component analysis are all included in the pre-processing of the signal.The blood pressure prediction findings for virtually all of the volunteers can fulfill the AAMI criteria after using the data from the trial to confirm this.This study develops a non-invasive blood pressure monitoring system that serves as inspiration for the creation of related intelligent wearable technology.