Automatic prediction of paroxysmal atrial fibrillation in patients with heart arrhythmia

Abstract

A new predictor that takes into accounts the randomness of RR interval before PAFib is proposed. Using data mining techniques, temporal patterns are identified based their presence in ECG that precedes paroxysmal atrial fibrillation (PAF) and not present in patients with normal ECG. The algorithm used the supposition that the premature atrial complexes (PAC) are responsible for most of PAF. Other statistical parameters that are related to randomness of signal are used to improve the accuracy of proposed algorithm.