基于支持向量回归的人体血压预测方法

Predicting method for human blood pressure based on SVR algorithm

针对当前基于机器学习的血压测量方法的准确率低下等问题,本文提出了一种基于支持向量回归的高效便捷的血压预测方法。算法首先分析人体生理指标数据与血压之间隐含的关系,建立血压预测模型,然后将算法结果与三个经典机器学习算法(线性回归、神经网络、岭回归)结果进行对比,利用两种评价指标(准确率、均方根误差)对它们进行评估,实验结果表明支持向量回归算法能准确高效地预测血压且优于其它算法模型。

In view of the problems of long timing for take measurement,the harm causing by continuous measurement to the body and the cumbersome measurement process,an efficient and convenient blood pressure prediction algorithm based on support vector machine regression algorithm was proposed. It firstly analyzed the implicit relationship between human physiological index data and blood pressure and then established the SVR Model. The results of the algorithm were compared with those obtained from three classical machine learning algorithms,i. e. linear regression model,ridge regression model,and neural network model,against two evaluation indexes（ accuracy,root mean square error）. The experimental results showed that support vector machine regression model（ SVR） can accurately and effectively predict blood pressure and be superior to other algorithms.