KardiaMobile detects atrial fibrillation (AFib) with a sensitivity between 86% and 99% and a specificity between 92% and 98%, depending on the study and reference standard used. That puts it in a remarkably useful range for a device you can hold in your hands, though it falls short of a full hospital-grade ECG in several important ways.
AFib Detection Accuracy
The strongest evidence comes from four peer-reviewed studies reviewed by the UK’s National Institute for Health and Care Excellence (NICE). Across those studies, KardiaMobile’s built-in algorithm detected AFib with 92% to 99% sensitivity and 92% to 98% specificity per recorded ECG. Sensitivity tells you how often the device correctly catches AFib when it’s actually present; specificity tells you how often it correctly identifies a normal rhythm as normal.
When compared head-to-head against a standard 12-lead hospital ECG in 98 cardiology patients, KardiaMobile 6L detected AFib with 86.4% sensitivity and 97.4% specificity. That means it missed roughly 1 in 7 AFib cases but very rarely called something AFib when it wasn’t. In the same study, every single AFib diagnosis made by KardiaMobile matched when reviewed by a physician, with 100% consistency across 21 confirmed cases.
These numbers are strong enough for screening and monitoring, but not strong enough to replace a clinical ECG for a definitive diagnosis. If KardiaMobile flags AFib, your doctor will almost certainly confirm it with a standard ECG before starting treatment.
Accuracy for Other Heart Rhythms
KardiaMobile is FDA-cleared to identify normal sinus rhythm, atrial fibrillation, bradycardia (slow heart rate), and tachycardia (fast heart rate). Its accuracy extends beyond AFib, though the evidence varies by condition.
In a study comparing single-lead ECG readings against cardiac electrophysiology testing (the most precise way to diagnose rhythm problems), accuracy for detecting premature ventricular contractions (extra heartbeats originating in the lower chambers) was 91%. Supraventricular tachycardia, a type of abnormally fast rhythm from the upper chambers, was detected with 89% accuracy. Ventricular tachycardia, a more dangerous fast rhythm from the lower chambers, reached 91% accuracy.
For sinus rhythm specifically, the device showed 88.1% sensitivity and 89.7% specificity when compared to a 12-lead ECG. That’s slightly lower than its AFib numbers, largely because conditions like atrial flutter or other borderline rhythms can look similar to normal sinus rhythm on a single-lead tracing.
What It Cannot Detect
KardiaMobile does not detect heart attacks. AliveCor’s labeling states this explicitly, and the FDA has documented adverse event reports tied to this misunderstanding. The device reads your heart’s electrical rhythm, not blood flow or muscle damage. A heart attack involves blocked blood supply to the heart muscle, which requires different diagnostic tools like blood tests and imaging. You could be having a heart attack and receive a “Normal” reading from KardiaMobile.
The device also cannot reliably detect all structural heart problems, valve disorders, or subtle conduction abnormalities that a full 12-lead ECG would catch. It’s a rhythm screening tool, not a comprehensive cardiac workup.
Six-Lead vs. Single-Lead Models
The KardiaMobile 6L captures six ECG leads instead of one, which gives physicians more information to work with. In a study of 176 adults with congenital heart disease, physicians using the 6L correctly classified the shape of the heart’s electrical signal (QRS morphology) in 90% of patients, significantly better than some competing devices.
The six-lead model also made it easier to identify key intervals on the ECG tracing. The PR interval, which reflects the timing of electrical conduction between the upper and lower chambers, was assessable in 89% of recordings. The QT interval, which matters for patients on certain medications, could be measured in 97% of recordings. These rates were substantially higher than competing single-lead devices.
However, the 6L’s built-in algorithm marked 31% of its own recordings as inconclusive, compared to just 5% for one competing device. This means the raw tracing quality is good for a doctor to review, but the automated interpretation struggles more often. If you’re using KardiaMobile primarily to get a quick algorithm-generated answer on your phone screen, the high inconclusive rate is worth knowing about.
QT Interval Monitoring
Some medications, particularly certain antibiotics, antipsychotics, and anti-arrhythmics, can dangerously prolong the QT interval, which reflects how long your heart takes to recharge between beats. A QT interval above 500 milliseconds is a red flag for potentially fatal rhythm disturbances.
A JAMA Network Open study of 191 patients found that at the critical 500-millisecond threshold, the KardiaMobile 6L had a negative predictive value of 99.8%. In practical terms, if the device says your QT interval is fine, it almost certainly is. However, the positive predictive value was only 16.7%, meaning most of its “prolonged QT” warnings turned out to be false alarms when checked against a standard ECG. The device tends to slightly overestimate QT duration, with average readings about 4 milliseconds longer than those from a 12-lead ECG.
This makes the 6L useful as a safety net for ruling out dangerous QT prolongation, but not reliable for confirming it without follow-up testing.
Factors That Affect Recording Quality
Even the best algorithm can only work with the signal it receives, and several common situations degrade KardiaMobile’s recordings:
- Dry skin or excess body hair reduces electrical contact with the electrodes. If your hands are very dry, applying a water-based lotion before recording helps.
- Movement during recording introduces noise into the tracing. Resting your arms on a flat surface while recording reduces muscle tremor artifacts.
- Electronic interference from charging cables, headphones, or nearby electronic equipment can distort the signal. Don’t record while your phone is charging.
- Dirty electrodes impair signal quality. The metal contact pads need occasional cleaning.
When the device can’t classify a recording, it labels the result “Unclassified,” meaning the tracing fell outside the algorithm’s standard boundaries. This isn’t necessarily a sign of a heart problem. It often just means the recording quality was poor. Retaking the recording with better technique usually resolves it.
How to Interpret Your Results
KardiaMobile’s accuracy is high enough to be genuinely useful for two main scenarios: catching intermittent AFib that might not show up during a brief office visit, and providing your doctor with a recording of what your heart was doing when you felt symptoms like palpitations or dizziness. The device captures a 30-second ECG tracing that you can share directly with your physician, and that tracing is often more valuable than the algorithm’s automated label.
The accuracy gap between the device and a hospital ECG matters most at the edges. KardiaMobile performs well for clear-cut cases of AFib and normal rhythm, but borderline or unusual rhythms are more likely to be misclassified or labeled unclassified. A study comparing devices in a cardiology clinic found that recording quality was rated “good” in 80% of KardiaMobile recordings versus 82% for the standard 12-lead ECG, a surprisingly small gap for a pocket-sized device.
Where the device consistently falls short is in precise measurements of ECG intervals. QT intervals were significantly shorter on KardiaMobile than on standard ECGs (366 vs. 403 milliseconds), and QRS duration showed small but statistically significant differences. These discrepancies matter for cardiologists making medication decisions but are less relevant for the typical user monitoring for AFib episodes.