Identification of common cardiovascular diseases based on the fusion of wrist pulse and electrocardiographic features
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 average precision of different models were calculated and comparison.Results:Wrist pulse features h2/h1,h3/h1,h5/h1,t1,t3,t1/T,t4/T,t5/t4,w1,w2,w1/T,w2/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 cardiovascular diseases.
NETL
NSTL
万方数据
