A neural network approach for predicting and modelling the dynamical behaviour of cardiac ventricular repolarisation

Physiological signals are usually patient specific, and they are difficult to predict, especially for the cardiovascular system. New methods capable to be adapted to each case and to learn the singular behavior of heart functions should be developed to support physicians in their decision-making. One of the most widely studied relations is the QT-RR one, between the total duration of the ventricle activation and inactivation, and the heart rate. In the past, different studies were made to approach this relation in the steady state. In this paper, a new method for modeling and predicting the transient dynamic behaviour of QT interval in relation to changing RR intervals is presented using artificial neural networks.