Research On Non-invasive Blood Pressure Measurement Method Based On Deep Learnin

Blood pressure(BP)is an important physiological index of human body,which can judge the cardiovascular function and heart condition of the body.Many diseases are closely related to blood pressure.Therefore,the correct determination of the specific value of blood pressure is of great significance for the diagnosis and treatment of cardiovascular diseases.At present,the main clinical use is invasive direct measurement,which has a high accuracy,but it is traumatic to the human body and is suitable for special people.Only noninvasive blood pressure measurement can be applied to daily life.In addition,blood pressure measurement based on pulse signal has become a hot topic.In this paper,the method of deep learning is mainly used to measure blood pressure based on photoelectric volume pulse wave,and the accuracy of blood pressure measurement is improved by improving the basic model.The main research work and conclusions of this paper are as follows:1.A noninvasive blood pressure measurement method based on long-term recursive convolution neural network is proposed.Firstly,the processed data is input into CNN neural network and LSTM neural network for experimental operation.According to the output results,the data values of the two networks are not much different,so we try to integrate the two models to form a new model-long-term recursive convolution network.Experiments using this network show that the experimental effect of the new model is better than CNN network and LSTM network.Compared with these two networks,the MSE error of the proposed method is reduced by 39.97% and 37.71% respectively,the MAE error is reduced by 14.06% and 17.29% respectively,and the R-square value is increased by 0.31% and 0.29%respectively.2.A noninvasive blood pressure measurement method based on BiLSTM network is proposed.Firstly,taking the BiLSTM network and the traditional LSTM as the experimental model,and comparing the output evaluation index coefficients,it is found that the BiLSTM network has a better effect on blood pressure measurement.Because the attention mechanism can assign weight coefficients from rows according to the importance of features,it is introduced into the BiLSTM network with good measurement effect for experiments.According to the results,it is found that compared with the original BiLSTM model,the MSE value and MAE values of the introduced attention mechanism model are reduced by 18.29%and 21.27% respectively,and the R-square value is increased by 0.17%.3.A non-invasive blood pressure measurement method based on CNN and BiLSTM network is proposed.This method combines the advantages of the first two models in improving the original structure,makes rational use of the advantages of CNN network in learning high-dimensional features,the feature that BiLSTM network can extract features in both directions and the feature that attention mechanism can allocate weights according to the criticality,and combines the three into a new model structure.The experimental results show that the model has a fast convergence speed in the training process,and the measurement of blood pressure is more accurate.Compared with the long-term recursive convolution network and the attention mechanism BiLSTM network,it is found that the measurement accuracy of this method is improved,the mean square error is reduced by 21.63% and 18.27%respectively,the average absolute error is reduced by 67.5% and 61.59% respectively,and the deterministic correlation coefficient is increased by 0.42% and 0.41% respectively.