Your watch vibrates at 2 a.m. and tells you your heart rhythm looks irregular. What now?
The AFib-detection feature on modern smartwatches is one of the biggest wins for consumer health tech in the last decade, and one of the most misunderstood. This is a walk-through of what a wearable ECG actually does, when an alert deserves your attention, and when it is probably a false alarm. We will link the two products in the Vyvata catalog with real cardiac clearance and be honest about what none of them can do.
What the sensor on your wrist actually sees
An ECG (electrocardiogram) records the electrical activity of the heart. A clinical 12-lead ECG places ten electrodes on your chest and limbs, capturing signal from twelve angles. The output is a picture of your heart from twelve viewpoints, which is why a cardiologist can identify not just rhythm problems but ischemia, chamber enlargement, and conduction blocks from it.
A smartwatch has one electrode against your wrist and asks you to touch a second electrode on the crown or bezel with a fingertip from the opposite hand. That closes the circuit and produces a single-lead ECG — one angle, roughly equivalent to lead I on a clinical tracing. It is genuinely useful for detecting some rhythm problems, particularly atrial fibrillation. It is not diagnostic for most other cardiac conditions.
The related feature, the passive AFib notification, uses a different sensor. That is your green-light photoplethysmography (PPG) sensor pulsing dozens of times per minute against your skin, watching for the irregular pulse pattern AFib usually produces. It runs in the background without you asking. When it sees irregularity for long enough, it prompts you to take an ECG to confirm.
The Apple Heart Study, and why the number 84 matters
In November 2019 the New England Journal of Medicine published the Apple Heart Study. The design: 419,297 self-enrolled Apple Watch users. Watches ran the irregular-pulse notification algorithm passively. Anyone who received an alert was mailed an ambulatory ECG patch to wear for up to a week, so investigators could confirm what the watch had flagged.
Of the participants who got a notification and returned patch data, roughly one in three had confirmed AFib on the gold-standard recording. The most-quoted number from the paper — a positive predictive value of about 84 percent — comes from the subset where the watch's algorithm concurrently observed irregular pulse tachograms alongside the notification. In that subset, AFib on the confirmatory patch was highly likely.
That is the strongest data any consumer wearable has ever produced for cardiac screening. It is also less impressive than it sounds if you skim it. A large fraction of participants who got a notification did not have AFib on the patch. That is not a failure of the algorithm; it is what happens when any screening tool is applied to a mostly healthy population. Which brings us to Bayes.
Why the same alert means different things for different people
The positive predictive value of a test — the probability that a positive result is real — depends on the underlying prevalence of the condition in the person being tested. This is not a design flaw. It is arithmetic.
Consider a 32-year-old with no risk factors and a resting heart rate of 55. Baseline AFib prevalence in that population is well under 1 percent. If the watch has 98 percent specificity, the positive predictive value of a single alert in that person may be only 20 to 30 percent — meaning most alerts in low-risk users are noise. Motion artifact, poor sensor contact from a loose band, high-intensity intervals, or a run of PVCs (premature ventricular contractions, mostly harmless) can all light up the algorithm.
Now consider a 72-year-old with hypertension, sleep apnea, and mild heart failure. Baseline AFib prevalence in that population is closer to 8 to 10 percent. The same watch, the same alert, in that person may carry a PPV well above 90 percent. That is the population the FDA framework was written for.
Practical translation: an alert is not a diagnosis. It is a prompt to look more carefully. The rate at which the alert is right depends on who you are.
510(k) vs de novo — what the FDA label actually means
The regulatory labeling here is worth understanding. The FDA has two main pathways relevant to wearable ECG features.
510(k) clearance means the manufacturer has demonstrated that the device is substantially equivalent to a device already on the market (called the predicate). For wearable ECGs, the predicate chain traces back to the AliveCor Kardia Mobile single-lead ECG, cleared years earlier. A 510(k) is a real regulatory review — it requires performance testing, software documentation, and cybersecurity review — but it is not the same as full de novo authorization.
De novo classification is what a genuinely novel device requires. Apple's original irregular-rhythm notification and Apple's ECG feature both went through de novo pathways in 2018 because there was no substantially equivalent predicate at the time. Withings and Fitbit followed with their own clearances. Most later entrants use 510(k) referencing those earlier de novo devices as predicates.
In practical terms, do not read "FDA cleared" as "FDA approved to diagnose AFib." The label the FDA actually authorizes is narrower — usually along the lines of "for the passive detection of possible irregular heart rhythm consistent with atrial fibrillation in adults, not intended as a substitute for clinical diagnosis." That wording is careful for a reason.
The two wearables with real cardiac clearance in our catalog
The ScanWatch 2 is the wearable we point most people toward when they want a device whose cardiac claims have been through the FDA. Withings publishes peer-reviewed validation work for its ECG classifier, which is unusual in the category. The 30-day battery matters more than it sounds — a watch that lives on the charger is a watch that is not on your wrist when the rhythm goes irregular.
The original Fitbit Sense and the Sense 2 both carry FDA clearance for their ECG feature and their irregular heart rhythm notification. Fitbit's algorithm has been through validation studies published in peer-reviewed journals and their disclosure of methodology is more complete than most of the wristband market. If you already use Fitbit, this is the cardiac-clearance path that keeps you inside the ecosystem you already trust.
The out-of-catalog benchmark: Kardia Mobile
We do not currently carry it, but for context it is worth naming: the AliveCor Kardia Mobile is the original consumer single-lead ECG. It is a small pad you press with your thumbs, and it produces a 30-second single-lead tracing a physician can read. The 6-lead version adds a knee electrode to give six leads instead of one, closer to what a clinician expects.
For someone who has already had an AFib episode, a family history of arrhythmia, or a cardiologist actively monitoring them, a dedicated Kardia device produces cleaner tracings and lets those tracings be sent directly to a cardiologist for review. It is a purpose-built cardiac tool rather than a watch that also does cardiac. When we add Kardia to the catalog we expect it to score in the Verified range.
What to actually do if your watch alerts you
The single most important thing to do when a wearable alerts you to a possible irregular rhythm is not panic, and not ignore. Both are common. Both are wrong.
- Take the ECG on the watch itself. If the passive notification triggered, run the manual single-lead ECG immediately. Sit still, arms supported, follow the prompt. Save the tracing to your phone.
- Note the timing and context. Were you exercising? Waking up? Right after alcohol? A stressful phone call? The trigger context matters for interpretation.
- Check for symptoms. Palpitations, chest pressure, shortness of breath, dizziness, or fainting are red-flag symptoms that change the situation. Chest pressure with radiation to the jaw or arm is not a "call your primary later" situation — it is a call-emergency-services situation.
- Schedule an appointment. Bring the tracing. A single wearable alert with no symptoms is almost never an emergency, but it warrants a proper 12-lead ECG in a clinical setting, and often a 24 to 72 hour Holter monitor or a 7 to 14 day patch monitor to characterize the rhythm across many hours.
- Do not stop your alerts. The temptation after a false alert is to disable notifications. If you are in a demographic where AFib matters, a false positive is a much smaller cost than a missed real event.
A note on wearing the watch to sleep
Passive AFib detection depends on time on skin. Most modern arrhythmia algorithms sample for several minutes at a stretch, then wait, then sample again — over the course of hours. A watch worn only during working hours will catch far fewer irregular episodes than one worn 24 hours a day, because AFib often shows up preferentially at night, when parasympathetic tone shifts. If cardiac screening is the reason you bought a wearable, wearing it to sleep is not optional. It is where a meaningful fraction of the value lives, which is another reason the ScanWatch's 30-day battery matters: it stays on the wrist through the night without a nightly charging ritual.
What the wearable ECG cannot see
To close the honest ledger, here is a list of cardiac conditions a single-lead wristwatch ECG will not detect reliably.
- Silent ischemia and heart attacks. The ST-segment changes that identify an acute coronary event show up on the leads that look at specific regions of the heart. Lead I alone will miss most of them.
- Ventricular arrhythmias. Ventricular tachycardia and ventricular fibrillation are the most dangerous rhythms, and they are not what a passive AFib detector is trained on. Wearables have flagged some, but they are not the target.
- Chamber enlargement, hypertrophy, and structural disease. These require an echocardiogram, not an ECG at all.
- Paroxysmal AFib that resolves before the watch samples it. Passive PPG monitoring is intermittent. Very brief episodes may not be caught. Symptoms plus a normal watch reading is still worth a workup.
The honest bottom line
A wearable AFib alert is one of the best things consumer health tech has ever produced. It has genuinely caught real AFib in real people who were not looking for it — the case reports are numerous and the epidemiology is now published. It has also cried wolf at hundreds of thousands of healthy adults who spent an afternoon anxious about nothing.
The right way to hold both truths at once: your watch is a screening tool with a well-characterized false-positive rate. It sits on your wrist for 30 days a month at essentially zero marginal cost. When it alerts you, get a real ECG. When it does not, do not confuse silence with a clean bill of cardiac health. A wearable is a starting point for a conversation with a physician, not a substitute for one.
If you want the version of that starting point with the deepest FDA paper trail and the longest battery, the Withings ScanWatch 2 at 85 Verified is where we would start. If you are already in the Fitbit ecosystem, the Sense at 80 Verified is the same category of tool with a different data home. Either one is a real screening device. Neither one is a cardiologist.