基于双路脉搏波的连续血压估算

Continuous Blood Pressure Estimation Based on Dual Pulse Waves

通过双路脉搏波传感器采集了掌部和手指的脉搏波数据,根据其提取了PTT特征和脉搏波波形特征。利用提取的特征训练基于遗传算法优化搜索参数的Xgboost模型。在对普通测试者进行早晚实验和运动前后对比实验数据进行分析对比发现,该模型预测得到血压平均绝对误差小于5mmHg,满足AAMI标准,在对高血压患者的验证实验中发现,其收缩压和舒张压92.5%的差异值处于一致性界限内。

The pulse wave data of palms and fingers are collected by a dual-channel pulse wave sensor,and the PTT characteristics and pulse wave waveform characteristics are extracted based on it.The extracted features are used to train the Xgboost model based on genetic algorithm to optimize the search parameters.In the morning and evening experiments on ordinary testers and the comparison of experimental data before and after exercise,it is found that the model predicts that the average absolute error of blood pressure is less than 5mmHg,which meets the AAMI standard.In the verification experiment on hypertensive patients,it is found that the systolic blood pressure and The diastolic pressure difference of 92.5%was within the limit of agreement.