2025年01期 v.46 1-13页
郭彬 杨其宇
（广东工业大学，广东 广州 510006）
摘要：为便于对高血压患者进行长期血压监测，提出一种基于多波长脉搏波的血压预测方法。首先，利用多波长脉搏波信号采集系统获取多波长脉搏波信号，并从中推导出脉搏传导时间；然后，结合多波长脉搏波的其他特征构建特征向量矩阵；并利用Lasso回归先对特征向量矩阵进行筛选，再进行血压预测。实验结果表明，该方法准确性较高，预测收缩压、舒张压的平均绝对误差分别为4.92、5.68 mmHg，标准差分别为6.15、6.68 mmHg，可作为血压测量的有效补充方法。
关键词：多波长脉搏波；血压预测；Lasso回归；脉搏传导时间
中图分类号：R318; TP181 文献标志码：A 文章编号：1674-2605(2025)01-0007-06
DOI：10.3969/j.issn.1674-2605.2025.01.007 开放获取
Blood Pressure Prediction Method Based on Multi-wavelength Pulse Wave
GUO Bin YANG Qiyu
(Guangdong University of Technology, Guangzhou 510006, China)
Abstract: To facilitate long-term blood pressure monitoring of hypertensive patients, a blood pressure prediction method based on multi wavelength pulse wave is proposed. Firstly, using a multi wavelength pulse wave signal acquisition system to obtain multi wavelength pulse wave signals, and deriving pulse transit time from them. Then, a feature vector matrix is constructed by combining other features of multi wavelength pulse wave, and use Lasso regression to screen the feature vector matrix first, and then perform blood pressure prediction. The experimental results show that this method has high accuracy, with average absolute errors of 4.92 and 5.68 mmHg in predicting systolic and diastolic blood pressure, and standard deviations of 6.15 and 6.68 mmHg, respectively. It can be used as an effective supplementary method for blood pressure measurement.
Keywords: multi-wavelength pulse wave; blood pressure prediction; Lasso regression; pulse transit time
