基于BP神经网络的血压监测算法

Blood pressure monitoring algorithm based on BP neural network

针对目前血压监测算法存在的不足,提出一种基于BP神经网络的血压监测算法。对光电容积脉搏波(PPG)、加速脉搏波(APG)及血压数据进行特征提取,得到10维输入向量和2维输出向量,将以上参数输入到BP神经网络中进行训练,得到血压监测算法模型的输出结果。经过计算平均绝对误差(MAE)、均方根误差(RMSE)以及通过线性回归分析评估该算法,评估结果表明,该算法具有良好的血压监测效果。此外,基于脉搏传感器MAX30102及STM32单片机设计硬件电路,采集测试者的脉搏波数据,并结合提出的算法进行实验。实验结果表明,该算法对收缩压和舒张压检测的误差分别为1.205±0.865 mmHg和1.3±0.5 mmHg,标准差分别为1.255±0.825 mmHg和1.465±0.515mmHg,符合AAMI血压测量标准,对血压实时监测具有较好的应用价值。

Aiming at the shortcomings of the existing blood pressure monitoring algorithm,a blood pressure monitoring algorithm based on BP neural network is proposed.The features of photoplethy⁃smography(PPG),accelerated photoplethysmography(APG)and blood pressure are extracted to obtain 10-dimensional input vector and 2-dimensional output vector.The parameters are input into BP neural network for training,and the output result of blood pressure monitoring algorithm model is obtained.Calculating the Mean Absolute Error(MAE),Root Mean Square Error(RMSE)and evaluating the algorithm through linear regression analysis,the evaluation results show that the algorithm has excellent blood pressure monitoring effect.In addition,the hardware circuit based on the pulse sensor MAX31002 and STM32 single chip microcomputer is designed to collect the pulse wave signal of the tester,and the experiment is carried out combined with the proposed algorithm.The experimental results show that the errors of the algorithm for the detection of systolic pressure and diastolic pressure are 1.205±0.865 mmHg and 1.3±0.5 mmHg respectively,and the standard deviations are 1.255±0.825 mmHg and 1.465±0.515 mmHg respectively,which meets the AAMI blood pressure measurement standard,and has great application value for real⁃time blood pressure monitoring.