The automatic detection and analysis of electrocardiogram based on Lorenz Plot

Lorenz Plot is widely used in analysis of heart rate variability because of its high sensitivity and high individuation. However, due to the high computational complexity and excessive reliance on experience based tuning, the effectiveness and accuracy of the traditional analysis methods based on Lorenz Plot are both low. In this paper, an automatic detection method according to the shape of Lorenz Plot is proposed, and a sequence of error analysis has been conducted according to the experimental results. First, the redundant points are automatically eliminated through coordinate transformation and their distribution characteristics along the X-axis and Y-axis. Then, we divide the coverage area of scattering points into several parts, and determine the shape of Lorenz Plot according to regional characteristics such as area scale, symmetry and slope. Finally, we conduct experiments with data from the MIT-BIH database, which prove that this method could quickly and accurately pick out the normal shapes from the complex shapes of Lorenz Plots. The detection accuracy of our method is above 90%, and the time complexity is O(n).