A Proposal for Cardiac Arrhythmia Classification using Complexity Measures

Abstract

Cardiovascular diseases are one of the major problems of humanity and therefore one of their component, arrhythmia detection and classification drawn an increased attention worldwide. The presence of randomness in discrete time series, like those arising in electrophysiology, is firmly connected with computational complexity measure. This connection can be used, for instance, in the analysis of RR-intervals of electrocardiographic (ECG) signal, coded as binary string, to detect and classify arrhythmia. Our approach uses three algorithms (Lempel-Ziv, Sample Entropy and T-Code) to compute the information complexity applied and a classification tree to detect 13 types of arrhythmia with encouraging results. To overcome the computational effort required for complexity calculus, a cloud computing solution with executable code deployment is also proposed.

Keywords

• complexity theory
• decision trees
• electrocardiography
• random sequences
• classification algorithms
• fuzzy set