Statistical Analysis Method Of Multi-lead ECG

This paper mainly introduced some statistical analysis methods to apply to the multi-lead ECG analysis. First we used a functional data analytic model to get four parameters from multi-lead ECG T waves; each set of four parameters represents one-lead ECG T wave. Further-more, we used mixed-effect models to estimate the effects of subject, lead and beat.Based on the functional data analytic model, we used principal component anal-ysis(PCA) and cluster analysis to reduce the numbers of variables, and obtained the different results using these two ways. Through applying PCA to the four parameters generated by the FDA approach, we found that the weight of each lead became similar, and the contribution of the first principal component increased by a large margin.Combining PCA, FDA and Monte Carlo simulation, this paper finally built model to fuse 12-lead ECGs. The simulation result showed that the variation of the fusion parameters had greatly reduced comparing with that of single lead parameters. So this method obviously improved the accuracy of automatic ECG analysis and can be used to analyze other high-dimensional signals.