[An algorithm study on telecardiogram diagnosis based on multivariate autoregressive model and two-lead ECG signals].

Objective: In view of the time delay caused by reconstruction of signals at remote sites, a direct classification method with high accuracy suitable for telediagnosis of electrocardiogram (ECG) signals is studied.

Method: The data for analysis and classification was obtained from MIT-BIH database, including 300 samples each of normal sinus rhythm (NSR), atria premature contraction (APC), premature ventricular contraction (PVC), ventricular tachycardia (VT), ventricular fibrillation (VF) and superventricular tachycardia (SVT). An multivariate autoregressive (MAR) model based technique that could combine the signals of two ECG leads was presented to classify the ECGs directly, including MAR modeling performed on ECGs, and quadratic discrimination function (QDF) based classification by using MAR coefficients and K-L MAR coefficients.

Result: Besides quick and convenient diagnosis, the accuracy of the proposed classification algorithm was as high as 98.3%-100%.

Conclusion: The MAR modeling based technique is suitable for telecardiogram diagnosis. Comparing with single-lead ECGs, better classification results can be obtained through the combination of two-lead ECG signals.