Detection of AF and I-AVB combined with RR interval and P wave

Atrial fibrillation (AF) and First-degree atrioventricular block (I-AVB) are two common arrhythmia diseases, which can lead to different cardiovascular diseases. Therefore, it is of great significance to realize the automatic diagnosis of AF and I-AVB. In this paper, a support vector machine (SVM) automatic detection method based on P-wave and RR interval features was proposed to classify normal (N), AF, and I-AVB. The method consists of three steps: 1) detection of the P wave and R wave reference points in the signal; 2) feature extraction, including analysis of P wave features extracted from the atrial activity and RR interval features extracted from ventricular activity; 3) detection using SVM classifier based on multi-classification strategy. First, the effectiveness of the feature combination was verified on the MIT-BIH AF database, in which the Acc rate was 92.40%, the Se was 96.25%, the Sp was 88.59%, and the F1 measure was 92.66%. The proposed algorithm was verified in China Physiological Signal Challenge 2018 database, and the results showed that the Se was 85.72%, the Sp was 92.38%, and the F1 measurement was 84.28%, with the Acc reaching 84.38%. It can be seen from the results that this method can effectively detect AF and I-AVB in the ECG signal.