基于脉象与心电特征融合的常见心血管疾病辨识研究  

Identification of common cardiovascular diseases based on the fusion of wrist pulse and electrocardiographic features

作　　者：武文杰 郭睿[1] 颜建军[2] 张春柯 王忆勤[1] 燕海霞[1] 马孝天 张叶青[3] WU Wenjie;GUO Rui;YAN Jianjun;ZHANG Chunke;WANG Yiqin;YAN Haixia;MA Xiaotian;ZHANG Yeqing(School of Traditional Chinese Medicine,Shanghai University of Traditional Chinese Medicine,Shanghai 201203,China;Institute of Intelligent Perception and Diagnosis,School of Mechanical and Power Engineering,East China University of Science and Technology,Shanghai 200237,China;Department of Cardiology,Shanghai Municipal Hospital of Traditional Chinese Medicine,Shanghai 200071,China)

机构地区：[1]上海中医药大学中医学院,上海201203 [2]华东理工大学机械与动力工程学院,上海200237 [3]上海市中医医院心内科,上海200071

出　　处：《中华中医药杂志》2024年第8期4010-4016,共7页China Journal of Traditional Chinese Medicine and Pharmacy

基　　金：国家自然科学基金项目(No.82074332);上海市科学技术委员会科研项目(No.19441901100)。

摘　　要：目的:探讨脉象与心电特征融合在常见心血管疾病辨识中的临床应用价值。方法:运用ZY-Ⅱ型压力-心电多源传感器脉诊仪同步采集常见心血管疾病患者(冠心病、高血压病、高血压病合并冠心病)的脉象信号和心电信号并提取特征参数,运用非参数检验法比较各组特征参数组间差异;基于不同的特征组合构建不同的建模数据集,运用随机森林算法建立常见心血管疾病的辨识模型,并计算准确率、精确率进行模型性能比较。结果:脉象特征h2/h1、h3/h1、h5/h1、t1、t3、t1/T、t4/T、t5/t4、w1、w2、w1/T、w2/T、P-RMSSD、P-SDNN和心电特征QRS时限、PR间期、心率差异存在统计学意义(P<0.05);基于脉象和心电特征融合的辨识模型,其准确率、平均精确率较基于脉象特征的模型、基于心电特征的模型分别提升了3.00%、2.81%和2.83%、2.89%。结论:脉象和心电特征融合可提高常见心血管疾病的辨识率。Objective:To explore the clinical application value of the fusion of pulse and electrocardiographic(ECG)features in the identification of common cardiovascular diseases.Methods:The zY-II pressure and echocardiogram multi-sensors pulse diagnostic device was used to synchronously collect pulse signal and ECG signal from patients with cardiovascular diseases(coronary heart disease,hypertension,and combined coronary heart disease and hypertension),and the features were extracted.Nonparametric test was used to compare the differences in features among the groups.Based on different features datasets,the random forest algorithm was used to establish identification models for common cardiovascular disease,and the accuracy and averageprecision ofdferent models were calculated and comparion.Results:Wrist pulsefeatures h/h,h/h,h/h,t,t3t/,t ts/t4,Wi,W2,w/T,w./T,P-RMSSD,P-SDNN and ECG features QRS duration,PR interval and heart rate were statistically different(P<0.05).The identification model based on the fusion of pulse features and ECG features showed improvements in accuracy,average precision compared to the models based on pulse waveform features or ECG features,with increase of 3.00%,2.81%,and 2.83%,2.89%,respectively.Conclusion:The fusion of pulse condition and ECG features can improve the identification rate of common cardiovasculardiseases.

关 键 词：心血管疾病辨识 脉象特征 心电特征 特征融合 随机森林 

分 类 号：R54[医药卫生—心血管疾病] TP181[医药卫生—内科学]