Classification of PVCs with a fuzzy logic system

The authors propose a fuzzy logic system for the classification of premature ventricular beats (PVCs) in ECGs. The classifier uses novel features extracted from a time-frequency analysis performed by a filter bank in addition to measurements related to the timing of R-R intervals and morphology analysis. The performance of the algorithm is evaluated on the MIT-BIH Arrhythmia Database following the AAMI recommendations. The achieved sensitivity is 81.34% and the positive predictivity is 80.64%.