In this paper, we proposed a heartbeat classification algorithm based on linear discriminant analysis and artificial neural network. For the input of classifier, we extracted 275 input features from the first derivative signal of ECG signal and RR interval information and it was reduced to be 6 by LDA. To evaluate the performance of the proposed algorithm, we compared the result of the proposed algorithm with that of fuzzy inference system classifier. MIT-BIH Arrhythmia database were used as test and learning data. The performance of the proposed algorithm was 97.49% for sensitivity, 97.91% for specificity and 96.36% for accuracy. For the extraction of features, the first derivative signal of ECG is used only so that the real-time implementation of this algorithm was possible. And, on account of the reduction of feature dimensionality, the time cost for learning and testing can be expected.

In this study, we proposed 17 input features based on wavelet coefficients for arrhythmia detection and, by applying linear discriminant analysis to these, reduced the feature dimension to be 4. Then, with newly constructed 4 dimension input feature, a multi-layer perceptrons classifier was tried to detect 6 types of arrhythmia beats. For evaluation of input features by linear discriminant analysis, the arrhythmia detection efficiency with these (LDA) was compared to that with original input features (ORG) and that with of input features by principle component analysis (PCA) respectively. When LDA was compared to ORG, the former showed similar or a little higher values than the latter for different types of arrhythmia beats except SVT. And, LDA showed to be persistently higher than PCA. By theses cross-validations, for the detection of several types of arrhythmia beats, the reduction of input feature dimension by linear discriminant analysis was revealed to be prior to that by principle component analysis. Even if LDA was compared to ORG, it maintained the acceptable level efficiency so that the time and computational costs would be expected to be cutdown dramatically. Finally, by the proposed algorithm, we could obtain the good accuracy of arrhythmia detection and that of NSR, SVT, PVC and VF was 99.52%, 99.43%, 98.59% and 99.88%, respectively.