Information transfer in QT-RR dynamics: Towards a model-free QT correction method

Heart dynamics is complex and results from interactions between various processes. The relation between the electrical properties of the heart and the beating rate is essential for the heart functioning. This relation is central when determining the corrected QT interval — an important measure of the risk for potentially lethal arrhythmias. We use the transfer entropy method from information theory to quantitatively study the mutual dynamics of the ventricular action potential duration (the QT interval) and the length of the beat-to-beat (RR) interval. This method allows for quantifying unidirectional information flows between the coupled processes and, thus, for assessing the degree of inter-dependence in an empirical data-driven manner. We show that the QT and RR intervals are coupled in a dynamical fashion. In particular, for healthy individuals there is a strong asymmetry in the information transfer: the information flow from RR to QT dominates over the opposite flow (from QT to RR), i.e., QT depends on RR to a larger extent than RR on QT. Moreover, the history of the intervals has a strong effect on the information transfer. For example, at sufficiently long QT history length the information flow asymmetry inverts, that is, the QT-to-RR transfer becomes larger than RR-to-QT and the RR influence on QT dynamics weakens. Additionally, we observe a critical history length of RR (about 25 heart beats), after which the RR-to-QT transfer no longer changes. Finally, we examine how the QT correction affects the information flows between QT and RR. We demonstrate that the current QT correction methods cannot properly capture the changes in the information flows between the coupled QT and RR time series. We conclude that our results obtained through a model-free information theory perspective can be directly utilized to significantly improve the present QT correction schemes in clinics.