Prediction of the behavior of the human heart using methods of non-linear dynamics

This thesis examines the behavior of the human heart from the point of view of non-linear dynamical systems theory. The non-linear analysis of ECG is performed by evaluating a number of parameters of interest in non-linear systems. The digitization of electrocardiograph recordings gathered from a group of patients by the Massachusetts Institute of Technology, Division of Health Sciences and Technology serves as the base for the analysis. Computer programs were developed to calculate the Correlation dimension, Lyapunov exponent, and the Kolmogorov entropy. These are calculated with the help of the algorithms developed by Grassberger and Procaccia (1983). These three quantitative methods are successful in proving the non-linearity of the system. The developed software implemented both qualitative and quantitative methods that help us to have a closer look at the behavior of heart. The values obtained for the three parameters are very comparable to those offered by Babloyantz and Dexteche [5] on a different data set. This tends to show that these parameters are universal parameters of the ECG. The software developed may help us to develop an innovative tool that may help us in long term behavioral analysis of human, which in the future may help in avoiding catastrophic events.