Regresar

Design of a low computational cost prototype for cardiac arrhythmia detection

Abstract:

This research presents the design of a prototype for the detection of cardiac arrhythmias that incorporates an embedded low-cost computational system in an environment of limited computational resources capable of analyzing characteristics of the QRS complexes. To do this, a strategy for classifying normal and pathological heart beats is developed in long-term electrocardiographic recordings (Holter), which are representative waves of the beat and their analysis allows identifying ventricular arrhythmias. For the development of this initial prototype, it is found that the use of the k nearest neighbors (k-NN) algorithm together with a stage of selection of variables from the training set is a good alternative and represents an important contribution of this work to experimental level. The experiments were carried out on the basis of cardiac arrhythmia data from the Massachusetts Institute of Technology (MIT). The results are satisfactory and promising.