Analysis the characteristics of ECG signals based on the transfer entropy

Transfer entropy reflects the varing trends of two signals. It can show the dynamic and directional information between two systems and can be applied to analysis nonlinear systems better than mutual information. In this article we apply transfer entropy to ECG signals, which extracts from the MIT-BIH ECG signal database. We compute the transfer entropy of ECG signals for different people and make comparison between the healthy and unhealthy group, and among different ages. Simulation results show that the value of transfer entropy for healthy people's ECG signals change little with the different sample length. The transfer entropy values of healthy people are increasing with ages when people's ages range from 20 to 35 years old, while they are gradually decreasing after 35 years old. The overall distribution of healthy people's transfer entropy is greater than that of non-healthy people. Under the case of illness people, the changing law of transfer entropy value is that the youth group is greater than the middle-aged group, and the middle-aged group is greater than the elderly group. The results of this paper have a certain significance in ECG signals analysis and detection.