基于支持向量回归模型的血压预测方法-《计算机技术与发展》

文章信息/Info

Title:: A Blood Pressure Prediction Method Based on Support Vector Regression Model

Author(s):: WU Xiao-jiao; WU Li-fa; School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China

Keywords:: blood pressure prediction; support vector regression algorithm; signal processing; sensor; photoplethysmography

摘要:: 血压实时检测对于及时了解人体的心血管系统状态具有重要意义。传统的侵入式和袖带法的血压测量方式是以间歇式为主,不能满足血压实时检测的需求。针对目前血压检测方式的不规范以及血压预测方法的准确度低下等问题,提出了一种仅使用光电容积脉搏波的基于支持向量回归模型的血压预测方法。该方法仅使用人的光电容积脉搏波生理信号,对该信号消除噪声污染和周期划分之后,再对原始的光电容积脉搏波信号以及其一阶导数和二阶导数提取相关特征,并使用支持向量回归算法构建预测血压的模型。基于 MIMIC III 数据库的数据进行实验,结果表明该模型能有效预测人的血压值,在均值误差和均方根误差方面的表现优于现有的方法,同时血压预测值大多数都在 96% 的一致性范围内。

Abstract:: Real-time detection of blood pressure is of great significance for timely understanding the state of the human cardiovascular system. The traditional invasive and cuff blood pressure measurement methods are mainly intermittent,which cannot meet the demand of real-time blood pressure measurement. In order to solve the problems of irregular blood pressure detection methods and low accuracy of blood pressure prediction methods,a new blood pressure prediction method based on support vector regression model using only photoplethysmography is proposed. In this method, only human photoplethysmography physiological signal is used. After eliminating noise pollution and period division? ? ?of the signal,relevant features are extracted from the original photoplethysmography signal as well as its first and second derivatives,and the support vector regression algorithm is used to build a model for predicting blood pressure. Experiments based on data from the MIMIC III database show that the model can effectively predict human blood pressure,and it performs better than existing methods in terms of mean? ? ? error and root mean square error. Meanwhile,the predictive values for blood pressure were mostly within the 96 percent consistent range.