Nonparametric Density-Based Clustering for Cardiac Arrhythmia Analysis

In this work, a nonsupervised algorithm for feature se-lection and a non-parametric density-based clustering al-gorithm are presented, whose density estimation is per-formed by Parzen’s window approach; this algorithm solves the problem that individual components of the mix-ture should be Gaussian. The method is applied to a set of recordings from MIT/BIH’s arrhythmia database with five groups of ar-rhythmias recommended by the AAMI. The heartbeats are characterized using prematurity in-dices, morphological and representation features, which are selected with the Q-α algorithm. The results are assessed by means supervised (Se, Sp, Sel) and non-supervised indices for each arrhythmia. The proposed sys-tem presents comparable results than other unsupervised methods of literature. 1