可穿戴设备的脉象特征分类及其结果的一致性分析

Classification of pulse characteristics of wearable devices and consistency analysis of their results

目的:基于可穿戴式脉象设备采集的脉搏波,探究不同方法对相同脉搏波分类的影响及分类结果的一致性。方法:对通过可穿戴式手环采集的脉搏信号进行了脉象特征提取与识别。首先对采集的脉搏波进行小波滤波去除噪声干扰,并归一化处理;然后分别使用基于时域特征的参数值法和基于欧氏距离的相似度法对脉搏信号进行分类;最后,采用Kappa检验对两种方法结果的一致性进行分析。结果:基于时域特征值参数法可以根据经验参数将脉搏信号进行程度等级区分,基于欧氏距离的相似度法可以通过测量脉搏波与标准脉搏波之间的相似度差异实现脉搏信号分类。两种脉象分类方法对脉搏波分类结果的Kappa值为0.42,表明分类结果一致性为中等。结论:结合不同的方法对脉搏信号进行识别将更有利于脉象客观分类的准确性。

Objective:To investigate the effects of different methods on the classification of the same pulse wave and the consistency of the classification results based on the pulse wave collected by the wearable pulse device.Methods:This paper extracted and identified the pulse characteristics of the pulse signal collected by the wearable bracelet.Firstly,the acquired pulse wave was wavelet filtered to remove noise interference and normalized.Then the pulse value signal was classified by the parameter value method based on time domain feature and the similarity method based on Euclidean distance.Finally,Kappa test was used.The consistency of the results of the two methods was analyzed.Results:Based on the time domain eigenvalue parameter method,the pulse signal could be graded according to the empirical parameters.The similarity method based on Euclidean distance can realize the classification of pulse signals by comparing the difference between the pulse waveform and the standard pulse waveform.The Kappa value of pulse wave classification results of the two pulse classification methods was 0.42,indicating that the consistency was moderate.Conclusion:Combining different methods to identify the pulse signal will be more conducive to the accuracy of objective classification of the pulse.