Nonlinear Analysis of ECG in Normal Subjects and Acute Myocardial Infarction Patients

Abstract

Background: In the last two decades, methods from nonlinear analysis and chaos theory have been used for the characterization of cardiac dynamics. Previous studies, mainly in the analysis of RR time series, indicated the diagnostic and prognostic value of such techniques in several pathological conditions, but also generated disputes regarding the presence of true nonlinearities in the data. The application of such methods to ECG time series analysis, in normal subjects as well as in patients with heart disease, is less investigated. The present study was undertaken to test whether nonlinear analysis of ECG can define possible changes in the complexity of cardiac dynamics in patients with acute myocardial infarction (AMI).Methods: We examined 10 normal subjects and 20 AMI patients in sinus rhythm for 3 consecutive days. Correlation dimension (D2) measurements were obtained with a combination of the SingularValue Decomposition and the Grassberger‐Procaccia methods and compared with the results from corresponding surrogate data series, in order to detect the presence of nonlinearities. The reproducibility of ECG D2 measurements was tested with the coefficient of variation.Results: No difference was found in ECG D2 between normals and patients. The surrogate data score was above two in all measurements, which indicates the presence of nonlinearities in the ECG signal. The intraindividual reproducibility of D2 measurements was moderate to good in normals and AMI patients.Conclusion: The results of this study support the hypothesis that the ECG signal is produced by a low‐dimensional dynamic system. D2 analysis of ECG is a nonlinear quantitative method with moderate to good reproducibility, but with no evident discriminating capabilities between normals and AMI patients.