Early diagnosis of atrial fibrillation (AF) can help get patients on therapy to prevent strokes, heart failure, urgent hospitalizations, and other complications. And mobile health technology now offers a variety of ways to screen for AF more effectively, including wearables like iRhythm’s Zio Patch or smartphone-paired ECG devices like the AliveCor Mobile ECG.

A study using the AliveCor Mobile ECG as an opportunistic screening tool in Australian pharmacy showed a detection rate of about 2% for undiagnosed AF. And several large studies are launching using the Zio patch to monitor at-risk patients for two weeks continuously to screen for AF.

Researchers from several institutions, including the University of Massachusetts and University of Connecticut, have developed an app that aims to skip the peripheral and use just the smartphone’s camera & flash to detect AF. We covered the original PULSESMART study in which they tested a prior version of this app

In this study, they tested an updated version of the app that aimed to address some of the key challenges they saw in their initial work: dealing with motion artifacts and discriminating frequent ectopy from atrial fibrillation.

To do that, they added 55 more participants to their cohort. Of these patients, 22 had AF and 33 had frequent ectopy (PVCs or PACs). Patients with AF were undergoing cardioversion so tracings were obtained both before (in AF) and after (in sinus). These patients were added to the existing dataset of 76 patients with AF undergoing cardioversion.

They found that the app had a 97% sensitivity and 94% specificity for detecting atrial fibrillation. Only two patients with ectopy were mischaracterized as having atrial fibrillation. They do note that sensitivity for PAC and PVC detection was lower (67-77%) so presumably the false negatives there were characterized as sinus rhythm.

There are some important advantages to an app-alone strategy for atrial fibrillation screening. First, it requires no equipment and can literally be downloaded anywhere, anytime by a patient. That could make atrial fibrillation screening incredibly scalable.

That said, the test characteristics in this study could be very different than when the app is out in the wild. First, the quality of the data, the pulse waveforms, probably won’t be as good as when data collection is being supervised by a study coordinator.

In addition, the prevalence of AF in the screened population will probably not be nearly as high, so that 94% specificity could translate into a lot more false positives and associated follow up testing. And here, that follow up testing is less likely to be a low-cost single-lead AliveCor Mobile ECG or even a 12-lead EKG in the office. More likely it’ll be some form of extended continuous monitoring to look for paroxysmal AF.

Finally, if cost trends in tech hold true, the cost of other screening strategies could head down over time. That could diminish the cost advantage of an app-alone screening strategy.

We’re likely to have abundant choices when it comes to screening for silent atrial fibrillation: an app, a smartphone peripheral, a wearable, or an implantable device.The challenge will be personalizing our screening strategies to match the right patient to the right screening modality, making sure that we start with each patient’s risk profile and fit the technology to that.