目前高血压已成为严重危害全球公共健康的重大问题.区别于传统的侵入式和袖带法的血压测量方式,为实时监测血压并助力早期诊断,本文专注于研究脉搏波波形与血压之间的内在关系,并提出 了 一种使用脉搏波的基于改进 BERT(Bidirectional encoder representationns from transformers)模型的血压预测方法.方法首先应用巴特沃斯滤波器对原始脉搏波信号进行滤波预处理并周期性划分,然后结合深度学习技术,采用改进后的BERT模型,对划分后的脉搏波周期数据进行特征提取和分析.为验证本方法预测的有效性和准确性,采用MIMIC-Ⅲ数据库的数据进行实验.实验结果表明,本方法可以有效预测血压值,完全满足英国高血压学会的A类标准.通过深入研究脉搏波与血压的关系,本文改进BERT模型为高血压的预测与诊断提供了新的技术手段.
Study on continuous blood pressure prediction method based on improved BERT model
High blood pressure has become a major issue seriously endangering global public health.Different from traditional invasive and cuff-based blood pressure measurement methods,this paper focuses on studying the intrinsic relationship between pulse waveforms and blood pressure.It proposes a blood pressure prediction method using pulse waves based on an improved BERT model.This method initially applies a Butterworth Filter for filtering and preprocessing the raw pulse wave signals and then divides them periodically.Combining deep learning technology,the improved BERT model is used for feature extraction and analysis of the divided pulse wave cycle data.To verify the effectiveness and accuracy of the proposed method,data from the MIMIC-Ⅲ database are used for experiments.The results show that this method can effectively predict blood pressure values,fully meeting the Class A standards of the British Hypertension Society.By thoroughly researching the relationship between pulse waves and blood pressure,this improved BERT model provides new technical means for the prediction and diagnosis of hypertension.
blood pressure predictionpulse waveButterworth filterimproved BERT model
万方数据
维普
