Abstract 9298: Detection of Atrial Fibrillation Using an iPhone 4S

Abstract

BACKGROUND: Atrial fibrillation (AF) is the most common sustained arrhythmia and is associated with increased risk for stroke, heart failure, and death. Timely detection of AF can be challenging using traditional diagnostic tools. Smartphone use is increasing among individuals susceptible to AF and may provide an inexpensive and user-friendly means to diagnose AF. Our aim was to to develop methods for accurately detecting AF using a smartphone camera. METHODS: In this prospective clinical investigation, we recruited 52 adults presenting to the University of Massachusetts Medical Center for electrical cardioversion of AF. Participants underwent rhythm assessment using a novel iPhone 4S application before and after electrical cardioversion. Two minutes of pulsatile time series recordings were obtained from each participant’s index finger (placed on the iPhone 4S camera). The iPhone 4S application used 2 statistical methods (Root Mean Square of Successive RR Differences; Shannon Entropy) to examine heart beat intervals and beat-to-beat variability. We used receiver operating characteristic (ROC) curves to establish a threshold for sensitivity, specificity, and predictive accuracy of our application as compared to the gold-standard of 12-lead electrocardiogram. RESULTS: The AF detection accuracy was 99.14% (sensitivity of 99.77% and specificity of 98.58%) for each sample when a combination of the three signal processing techniques were applied compared to the gold-standard electrocardiographic recordings. All of the 52 AF episodes were detected. In addition, we tested on 29 healthy subjects and found that the accuracy (specificity) was 99.51%. CONCLUSIONS: Our iPhone 4S AF Detection application performed well, demonstrating excellent sensitivity, specificity, and accuracy for beat-to-beat discrimination of AF from normal sinus rhythm. For clinical applications, the relevant objective is to detect the presence of AF from a given recording. Using this criterion, the AF detection accuracy was 100% based on a 2-minute sample. Our findings show that AF can be accurately detected from pulsatile signals in the human fingertip using the camera of an iPhone 4S.