Predicting the Percentage of Atrial Fibrillation using Sample Entropy

Abstract

Atrial fibrillation is the most commonly confronted cardiac arrhythmia in humans. This paper is written to use sample entropy and percentage of atrial fibrillation as a measure of regularity to measure AF. To assume the percentage of AF, 25 long term ECG recordings of human subjects with atrial fibrillation containing a total of 299 AF episodes were processed. The mean and SD of percentage breaking point in all the subjects from the MIT-BIH Atrial Fibrillation database was 0.6057±0.0863, and its sample entropy is 0.3522±0.1509. The mean and SD for sample entropy at 100% AF is 1.0669±0.4521. This data is used to predict the percentage of AF at a given sample entropy value. Our study concludes that the early detection of AF can be initiated by the AF already happened for 60%.