基于光电信号生理参数预测研究综述
A Review of Prediction of Physiological Parameters Based on Photoplethysmography
陈宇斌 1崔玉红 1梁启军 2邓皓明1
作者信息
- 1. 南昌航空大学 软件学院,江西 南昌 330063;物联网与大数据实验室,江西 南昌 330063
- 2. 江西中医药大学附属医院肺病科,江西 南昌 330006
- 折叠
摘要
由于光电信号(PPG)传感器具有身形小巧、佩戴方便等特点,基于其的研究备受欢迎.该文旨在阐述PPG信号在生理参数预测中的应用价值,并介绍传统机器学习算法和深度学习算法在各种生理参数上的研究进展.PPG采集的便利性将有助于多场所健康监测和疾病预防的应用.通过总结近年来PPG在各种生理参数估计方面的研究成果,提出了不同的生理参数的估计算法,推动了诊断方式的发展.主要从3 个方面展开:首先,对包含PPG的现有数据集进行整理,以展示不同数据集的信号数量以及所包含的其他种类信号,帮助研究人员查找和利用数据集;其次,对数据预处理方式进行概括,分析了不同预处理方法的优缺点,并提出改进方法以减少PPG信号在采集过程中受到的外界干扰;最后,对不同生理参数的预测算法进行了比较和分析,分模块概括介绍了不同生理参数的预测算法.
Abstract
The research based on photoplethysmography(PPG)sensors is particularly well-liked due to its properties,such as small size and ease of wear.We seek to demonstrate the usefulness of PPG signal in the prediction of physiological parameters and introduce the re-search progress of traditional machine learning algorithms and deep learning algorithms on various physiological parameters.The use of multi-site disease prevention and health monitoring will be made easier thanks to the convenience of PPG collecting.We provide several physiological parameter estimation algorithms by summarizing the research findings of PPG in recent years,which support the advancement of diagnostic techniques.This paper is mainly carried out from three aspects.Firstly,the existing datasets containing PPG are sorted out to show the number of signals in different datasets and other types of signals included,to help researchers find and use datasets.Secondly,the data preprocessing methods are summarized,the advantages and disadvantages of different preprocessing methods are analyzed,and an improved method is proposed to reduce the external interference received by the PPG signal during the acquisition process.Finally,the prediction algorithms of different physiological parameters are compared,analyzed and summarized by sub-mod-ules.
关键词
PPG/机器学习/生理参数/健康监测/预测Key words
photoplethysmography/machine learning/physiological parameter/health monitoring/prediction引用本文复制引用
基金项目
国家自然科学基金地区项目(62062050)
出版年
2023
国家科技期刊平台
NSTL
维普
万方数据