Automated home monitoring of atrial fibrillation in heart failure patients

Abstract

Atrial fibrillation (AF) is a major cause of cardiovascular complications. AF is likely to occur in heart failure (HF) patients and may precipitate episodes of worsening HF. AF is often underdiagnosed or detected late using traditional diagnostic tools. The increasing use of home telemonitoring (HTM) in the long-term management of chronic HF may facilitate the detection of incident and recurrent AF. Many HTM solutions for HF include devices that allow patients to take and transmit their daily heart rate along with other vital signs. Software-based processing of the cardiac signal underlying the heart rate assessment may reveal the presence of AF. In this study, we investigated the feasibility of detecting AF from the raw signal of a HTM device. The device records a short, non-conventional ECG to determine average heart rate. An algorithm based on Markov modelling of inter-beat-intervals (IBI) was used to distinguish AF from other heart rhythms. The approach was evaluated using daily self-assessments (n=3831) transmitted by HF patients over the course of one year. Implantable devices and patients' medical records served as a reference. On this dataset, the algorithm obtained a sensitivity of 94% and a specificity of 99% in discriminating AF from non-AF rhythms. Further studies should investigate the contribution of AF monitoring to the early detection of HF worsening.