Paroxysmal Atrial Fibrillation Screening at Different ECG Sampling Frequencies using Probabilistic Symbolic Pattern Recognition

Abstract

A myriad of data produced in intensive care unit (ICU) poses challenges in real-time processing and data storage. It is important to know the minimal sampling frequency requirement to develop real-time analysis on ICU data. In this study, we have applied a novel Probabilistic Symbolic Pattern Recognition method to screen subjects prone to paroxysmal atrial fibrillation by analyzing their electrocardiograph signals at different sampling frequencies varying from 128 Hz to 8 Hz. Results show that using the proposed method, we can obtain a classification accuracy of 82.6% in screening paroxysmal atrial fibrillation even when data is sampled at 8 Hz.