Your Apple Watch Flagged AFib — How Worried Should You Be?

Your Apple Watch buzzed at 2:14 PM on a Tuesday. You looked down: irregular rhythm detected. You weren't feeling anything — no palpitations, no shortness of breath, no chest pressure — but the notification was there, and it said the word "atrial fibrillation." You spent the rest of the afternoon cycling through the same loop: maybe it's nothing, maybe it's serious, maybe I should go to the ER, maybe I'm overreacting.

The Apple Watch's irregular-rhythm notification is one of the most sensitive screening tools ever put on a wrist. In a controlled lab setting against a 12-lead ECG, it catches about 96% of real atrial fibrillation when AFib is present. That sounds like a triumph of consumer health tech. And in some ways it is. But sensitivity is only one side of the equation. The other side — specificity — is where the story gets complicated, and where the notification's usefulness depends almost entirely on who you are.

The Base Rate Problem, Plainly

Atrial fibrillation is not evenly distributed across the population. In healthy adults under 50 — the kind of people who wear an Apple Watch to track their lifting sessions and sleep — the prevalence of AFib is roughly 0.5% to 1%. In adults over 65 with hypertension or diabetes, that number climbs to 5% to 10%. The same watch, running the same algorithm, produces a very different probability of being correct depending on which group you're in.

This is the base rate problem. A test that is 96% sensitive and 88% specific — numbers consistent with the 2025 diagnostic meta-analysis of smartwatch AFib detection — applied to a population with 0.5% prevalence means that for every 10,000 people, 48 cases of real AFib will be caught, but 1,194 people will receive a false positive. The positive predictive value — the chance that a positive notification is actually AFib — is about 4%. Ninety-six percent of the time, the notification is wrong.

In a 75-year-old population with 5% prevalence, the same arithmetic gives a PPV of roughly 40%. Same watch, same algorithm, same notification — but the actionability of that alert is an order of magnitude higher. The watch didn't get smarter. The base rate changed.

Why the Watch Gets It Wrong

The irregular-rhythm notification works by measuring pulse-to-pulse intervals from the photoplethysmography sensor on the wrist. It looks for the kind of irregularity that AFib produces — the classic "irregularly irregular" pattern — and flags it. But that pattern isn't unique to AFib. Several other phenomena produce the same signal.

Premature atrial and ventricular contractions — PACs and PVCs — are the most common cause of false positives. These are benign in young healthy hearts but produce an early beat followed by a compensatory pause, which the algorithm reads as a disrupted rhythm. Cardio-respiratory coupling during deep breathing can also raise the RMSSD — a measure of heart rate variability — high enough to trip the detector. Motion artifact from the wrist moving while the sensor is still trying to read intervals adds noise. Even caffeine combined with dehydration can increase ectopic beat frequency and produce a false alert.

None of these are dangerous. But they all look like AFib to an algorithm that's optimizing for sensitivity. The watch is designed to catch every possible case, and that design choice — the same one that gives it 96% sensitivity — is the same one that produces the false-positive rate.

What You Should Actually Do

The answer depends on who you are, not just what the watch said. If you're under 50, healthy, have no symptoms, and the alert is a one-off — the probability that this is a false positive is high enough that you don't need to go to the ER. You should not ignore it entirely. But you should treat it as data, not a diagnosis. Make an appointment with a primary care doctor, mention the alert, and get a real 12-lead ECG. The false-positive rate is high, but the cost of a confirmatory ECG is low, and the consequences of missing a real case are serious.

If you're over 65, especially with risk factors like hypertension, diabetes, or sleep apnea, the same alert carries a meaningfully higher probability of being real. You should schedule an ECG sooner rather than later, and you should be more willing to treat the alert as a signal that something needs investigation.

If the alert repeats — three times in a month, or five times in a quarter — that's a different situation. One alert is noise; a pattern is signal. The false-positive rate for a single alert is high, but the false-positive rate for repeated alerts drops because the algorithm is seeing the same irregularity across multiple days. That pattern should be taken seriously regardless of age.

The Broader Wearable Literacy Problem

The AFib notification is a specific case of a general problem: wearables generate alerts at a rate that exceeds the user's ability to interpret them. The same watch that flags an irregular rhythm also tracks your HRV, your resting heart rate, your sleep stages, and your blood oxygen. Each of those metrics has its own base rate problem, its own false-positive rate, and its own set of confounders. The Whoop memory problem — where a wearable stores a reading without the context that produced it — is the same issue in a different package. The Apple Watch numbers that change training are more reliable because they're trend-based rather than event-based, and trends are less vulnerable to the base rate problem than single-point alerts.

An HRV trend across two weeks is a signal. A single HRV reading is noise. A single AFib notification is noise, unless you're in a high-prevalence group. Understanding the difference between a signal and noise is the core of wearable literacy, and it's a skill that most users don't have because the device manufacturers don't teach it. The watch says "irregular rhythm." The watch doesn't say "this has a 4% chance of being real in a 30-year-old." That number is left for you to figure out, and most people don't.

When the Notification Is Useful

The AFib notification has real value for certain groups. In older adults with risk factors, the watch can catch paroxysmal AFib — the kind that comes and goes — that might not show up on a single office ECG. The Apple Heart Study, published in the New England Journal of Medicine in 2019, found that the watch's irregular-rhythm notification had a positive predictive value of about 71% in the subset of participants who received a notification and then wore an ECG patch. That study population was older on average — median age 41, but with a wide range — and the PPV was higher because the prevalence of AFib in that population was higher than in a general 30-year-old cohort.

For a 65-year-old with hypertension, the watch is a useful screening tool. For a 30-year-old who lifts four times a week and has no cardiac history, the watch is a source of anxiety-producing noise. The same device, the same algorithm, the same notification — but the medical relevance flips entirely based on the wearer's baseline risk.

This is not a failure of the technology. It's a failure of the framing. The watch can't know your age, your blood pressure, your family history, or your symptom status. It can only measure pulse intervals and apply a threshold. The interpretation is yours. And most people don't have the Bayesian reasoning to do that interpretation correctly.

The Short Version

A single irregular-rhythm notification on an Apple Watch in a healthy adult under 50 has roughly a 4% chance of being real AFib. The other 96% is a false positive from a PAC, a PVC, deep breathing, motion artifact, or a combination of caffeine and dehydration. You should not ignore it, but you should not panic. Schedule an ECG, mention the alert, and move on with your training. If the alert repeats, take it more seriously. If you're over 65 or have risk factors, treat the first alert as a meaningful signal.

The broader lesson is that wearables generate alerts that require interpretation, and the interpretation depends on who you are. The HRV higher not always better insight applies here too: context is everything. A number without a base rate is a headline without a story. The watch gave you a data point. The rest is on you.

Sources

The accuracy numbers in this piece draw from three bodies of work: Seshadri et al.'s 2020 lab validation in Circulation establishing the 96% sensitivity and 81% specificity of the Apple Watch against 12-lead ECG; the 2025 multi-wearable comparison in JMIR Formative Research that replicated those numbers across devices and quantified false-positive rates in screening populations; and the 2025 systematic review and diagnostic meta-analysis (PMC12713314) that pooled sensitivity at ~93% and specificity at ~88% and provided the Bayesian context showing that positive predictive value drops to roughly 20-30% at typical asymptomatic-young-adult prevalence — and to 4% when prevalence is 0.5%. The Apple Heart Study's 2019 NEJM results are the clinical anchor for the notification's real-world performance in higher-prevalence groups.