Skip to Main Content
Risk stratification for sudden cardiac death is one of the most compelling challenges of modern cardiology. An ECG-derived method of assessment of Heart Rate Variability (HRV) is one of few diagnostic modalities for that purpose whose applicability has been widely acknowledged. HRV reflects the mutual balance between the two branches of the autonomic nervous system (ANS), i.e. sympathetic and parasympathetic. In healthy individuals HRV remains high, while in diseased ones (e.g. presenting with severe heart failure) it becomes markedly reduced, indicating loss of equilibrium at the level of the ANS branches. It has been established that this reduction identifies patients with high risk for arrhythmic sudden cardiac death. Contemporary methods for measurement of HRV usually employ the Holter ECG recording [I] where a portable device continuously monitors the electrical activity of the heart for 24 hours or longer. The method has, however, at least two serious drawbacks. First, long duration of the ECG registration makes it less available for most people. Second, it renders the tracing to artifacts related to changes in heart rate due to wide range of activities and emotional stress experienced in every day life. Therefore, we believe that efforts should be made to develop a new method of HRV assessment using much shorter ECG signal recorded in conditions where effect of random, uncontrolled external stimuli and nonstationarites due to physiological processes would be minimized and/or standardized. To achieve this goal we introduce a novel method of HRV measurement employing nonlinear-dynamical (NLD) analysis of 30-minute ECG recording. Our approach is based on: (i) standardization of controlled external stimuli (Pilot-I and Pilot-2), (ii) introducing novel NLD complexity measures of the stimuli-perturbed ECG signal and (Hi) applying supervised classification methods as a step to incorporate our framework into a diagnostic decision support system. A brief description o- - f each of the above parts follows.