Research On Noninvasive Continuous Blood Pressure Measurement Based On GA-Elman Neural Network

Blood pressure is an important physiological index reflecting the function of heart and blood vessels.It is also an important basis for clinicians to diagnose diseases,evaluate therapeutic effects and judge prognosis.Blood pressure is susceptible to external and internal factors such as personal emotions,physiological cycles and so on.There are great differences in single measurement results.In daily life,continuous blood pressure measurement is of great significance.At present,continuous blood pressure measurement methods are mainly divided into non-invasive and invasive.The invasive method is easy to infect the subjects and is not suitable for daily life.Non-invasive method mainly refers to pulse wave conduction velocity method,pulse wave conduction time method and pulse wave characteristics method.Because the factors affecting blood flow are very complex,the measurement results of the first two methods are not accurate enough.With the vigorous development of artificial intelligence,the algorithm of predicting and estimating data by creating models has been gradually improved,and the measurement of continuous blood pressure by using pulse wave characteristic parameters has become a hot research direction.This paper mainly studies the non-invasive continuous blood pressure measurement method based on Elman neural network optimized by genetic algorithm.This method has simple structure and can get rid of the constraints of inflatable cuff in traditional methods.The main contents of this paper include:1.Remove the noise of pulse signal.Pulse wave has low frequency characteristics and is vulnerable to noise interference.Wavelet transform is used to remove baseline drift and high frequency noise of pulse wave.2.Differential threshold method is used to extract pulse wave characteristics.It can automatically recognize and extract the three characteristic points of pulse wave:the starting point,the peak point and the descending isthmus.Pulse wave characteristic matrix is established to prepare data for the follow-up study.3.Establish GA-Elman neural network continuous blood pressure measurementmodel.The synchronous pulse signal and continuous blood pressure data in MIMIC database are used as research objects.After signal pretreatment,the characteristic matrix is obtained to form the network input signal.Through many experiments,the genetic algorithm parameters are set,the population size and iteration times of the genetic algorithm are adjusted,the crossover and mutation probability are determined,and the fitness function is determined.Finally,A continuous blood pressure measurement model based on GA-Elman neural network was established.4.The optimization effect of genetic algorithm on Elman neural network and the accuracy of GA-Elman neural network continuous blood pressure measurement model are verified by comparative experiments.BP neural network and Elman neural network are selected to compare with the method in this paper.The experimental results show that the performance of GA-Elman neural network is better than that of BP neural network and non-optimized Elman neural network.A large number of data in MIMIC database are tested and the measurement errors of the three methods are counted.The experimental results show that GA-Elman neural network is superior to BP neural network and Elman neural network in accuracy and stability of blood pressure measurement.Elman neural network optimized by genetic algorithm has faster convergence and stronger learning ability.5.The continuous blood pressure measurement system based on GA-Elman neural network is preliminarily realized.The algorithm is implemented in Java and embedded in the host computer.The mobile client can query the blood pressure measurement results through the network.In the follow-up research work,the system can be further optimized to realize the practical application of GA-Elman neural network in continuous blood pressure measurement.