Evaluating smartphone based PPG as a screening solution for atrial fibrillation: a digital tool to detect AF?

Abstract presentation ACC 2016

October 6th, 2016

JACC, Vol 69, Issue 11, p2499

Objectives:

Guidelines indicate opportunistic screening for Atrial Fibrillation (AF) as a class I recommendation since it is proven to be important and effective in identifying cases of untreated, frequently asymptomatic AF.  This work focuses on comparing the performance between photopletyshmography (PPG) and single lead ECG based smartphone applications during a national incentivized screening initiative.

Methods

Participants presented themselves voluntarily at two screening sites during the screening initiative. Screening was done using sequential measurements a single lead ECG device (Alivecor, 30 seconds) and a software only smartphone application based on photoplethysmography  (FibriCheck, 60 seconds). Alivecor measurements were performed by placing both hands on two electrodes while the FibriCheck requires to place the finger on the smartphone camera. Additionally, demographic and background questionnaires were obtained. If one of the screening technologies indicated an irregular rhythm a 12-lead ECG was taken for verification by the cardiologist on site.

Results

In total 1056 participants were screened, 41% was male. The overall mean age was 59±15 years with a mean BMI of 26±10. In total 31% had no risk factors for AF, 34% had 1 risk factor, 19% had 2 risk factors and 16% had two or more risk factors. The screening resulted in the identification of 8 AF cases, 1026 regular sinus rhythms and 22 irregular rhythms (bigeminy, trigeminy, supraventricular arrhythmia).  The AF cases had a CHADS2-VASc score of 3±1.18. The Alivecor had a sensitivity of 100% and specificity of 99.6% for the detection of AF, while the  FibriCheck application had a sensitivity of 100% and a specificity of 97.2. Overall quality of the FibriCheck and Alivecor measurements was automatically determined and was unreadable in 2.9% and 3.8% of the cases respectively. These cases required an additional measurement to obtain a diagnosis. No correlation was found between the cases with bad quality measurements for both measurement techniques or demographics.  

Conclusion

The obtained results indicate that detection of pulse intervals based on PPG is a sensitive and accurate screening tool for the detection of atrial fibrillation and has a high level of agreement with the results obtained using the single lead ECG. The use of a smartphone-only application could unlock the potential of digital screening and support case finding of atrial fibrillation in selected population at risk for atrial fibrillation.

Participating Centers:

a Department of Cardiology, Ziekenhuis Oost-Limburg, 3600 Genk, Belgium.
b Mobile Health Unit, Faculty of Medicine and Life Sciences, Hasselt University, 3500 Hasselt, Belgium.
c Qompium NV, Wommelgem, België
d Department of Cardiology, AZ Delta, 8800 Roeselare, Belgium