Can Wearables Detect AFib? Clinical Evidence & FDA-Cleared Devices

Yes, modern smartwatches can detect atrial fibrillation with clinically meaningful accuracy, but they are screening tools rather than diagnostic devices. The Apple Heart Study demonstrated an 84% positive predictive value for irregular rhythm notification in over 419,000 participants, and multiple smartwatches now carry FDA clearance for ECG-based AFib classification. However, the distinction between screening (flagging potential cases for further evaluation) and diagnosis (confirming a medical condition with certainty) is critical for understanding what wearables can and cannot do.

Atrial fibrillation affects an estimated 33 million people worldwide and is a leading cause of stroke. Because AFib is often intermittent and asymptomatic, many people have it without knowing. Wearable technology offers a unique opportunity for long-term passive screening that could identify these individuals before a stroke occurs. This article reviews the clinical evidence, explains the technology, and discusses the limitations that both clinicians and consumers should understand.

For background on how the optical sensors in these devices work, see our guide to PPG technology.

How PPG Detects Atrial Fibrillation

PPG-based AFib detection does not directly measure the heart's electrical activity. Instead, it analyzes the timing of pulse waves arriving at the wrist (or finger) to infer cardiac rhythm regularity. During normal sinus rhythm, the intervals between successive heartbeats (pulse-to-pulse intervals, analogous to R-R intervals in ECG) show a characteristic pattern of variability modulated by respiration and autonomic tone. During atrial fibrillation, the chaotic electrical activity of the atria produces irregularly irregular ventricular contractions, meaning the pulse intervals become random with no discernible pattern.

The key algorithmic features used to distinguish AFib from sinus rhythm include:

• Root mean square of successive differences (RMSSD) of pulse intervals, which increases during AFib
• Shannon entropy of the pulse interval distribution, which increases with the randomness of AFib
• Poincare plot analysis, where successive interval pairs scatter broadly during AFib instead of clustering along the identity line
• Absence of P-wave equivalent features in the PPG waveform morphology (though this is more subtle and less reliable than interval analysis)
• Turning point ratio and other nonlinear metrics that quantify the irregularity pattern

These features are typically fed into machine learning classifiers (random forests, gradient boosting, or neural networks) trained on large datasets of labeled PPG recordings with simultaneous ECG verification. The algorithms are specifically tuned to distinguish AFib from other causes of irregular pulse intervals, particularly premature atrial contractions (PACs) and premature ventricular contractions (PVCs), which are far more common and generally benign. For details on the signal processing algorithms underlying these methods, see our technical guide.

The Apple Heart Study: Landmark Evidence

The Apple Heart Study (Perez et al., 2019, New England Journal of Medicine) remains the most influential study on wearable AFib detection. This prospective, single-arm study enrolled 419,297 Apple Watch users over eight months, making it one of the largest cardiac studies ever conducted.

Study Design and Results

Participants wore their Apple Watches normally, and the device's irregular pulse notification algorithm passively monitored pulse intervals using the green LED PPG sensor. When the algorithm detected a pattern consistent with AFib on multiple checks (at least five out of six tachograms over a minimum 65-minute period), the participant received a notification and was offered a telemedicine consultation and an ECG patch for confirmatory monitoring.

Key findings included:

• 0.52% of participants received an irregular pulse notification (2,161 of 419,297). Among participants over 65, the notification rate was 3.2%, consistent with known AFib prevalence in this age group.
• 84% positive predictive value for AFib detection when the irregular pulse notification was compared to simultaneous ECG patch recordings.
• 34% of notified participants who wore the ECG patch had AFib confirmed on the patch recording. This seemingly lower number reflects the intermittent nature of paroxysmal AFib; many participants were not in AFib during the patch monitoring period but did have it when the watch detected it.
• 57% of notified participants sought medical attention outside the study, indicating that the notifications prompted healthcare engagement.

Limitations of the Study

The Apple Heart Study had significant limitations that temper its conclusions. It was not a randomized controlled trial and had no control group. The study population was self-selected (Apple Watch owners who enrolled in a research app), skewing younger and healthier than the general population at highest risk for AFib. There was no systematic follow-up for false negatives (participants who had AFib but were never notified). The confirmatory ECG patches were worn for only about one week, potentially missing intermittent AFib in notified participants.

Despite these limitations, the study established that large-scale passive AFib screening using consumer wearables is feasible and produces clinically actionable results. It was instrumental in building the regulatory and clinical framework for wearable cardiac screening. For a broader perspective on wearable accuracy, see our Apple Watch versus chest strap accuracy comparison.

The Fitbit Heart Study and Other Large Trials

Fitbit conducted a similar large-scale study (the Fitbit Heart Study) using PPG-based irregular heart rhythm notifications. Published findings showed:

• Over 455,000 participants enrolled
• Positive predictive value of approximately 98% for AFib detection among those who received notifications and completed confirmatory ECG patch monitoring
• The higher PPV compared to the Apple Heart Study may reflect differences in algorithm tuning, study design, or population characteristics

Samsung, Google (Pixel Watch), and Withings have conducted or are conducting their own validation studies for PPG-based AFib detection, with Samsung's Irregular Heart Rhythm Notification receiving FDA clearance in 2024.

The convergence of results across multiple manufacturers and studies strengthens the evidence that PPG-based passive AFib screening is reliable. However, each device's specific algorithm, notification threshold, and clinical validation data should be evaluated individually. Not all wearables with heart rate sensors have validated AFib detection capabilities. Checking for FDA clearance status is an important step before relying on any device for health screening.

FDA-Cleared Devices for AFib Detection

As of early 2026, multiple wearable devices have received FDA clearance for AFib-related features. These clearances fall into two categories:

ECG-Based AFib Classification (De Novo 510(k))

• Apple Watch Series 4 and later: FDA-cleared ECG app that classifies 30-second single-lead (Lead I) ECG recordings as sinus rhythm, AFib, or inconclusive. Reported sensitivity 98.3%, specificity 99.6% for classifiable recordings.
• Samsung Galaxy Watch 3 and later: FDA-cleared ECG feature with similar single-lead classification capability.
• Withings ScanWatch: FDA-cleared for both ECG-based AFib detection and medical-grade SpO2 measurement.
• Google Pixel Watch: ECG AFib detection capability with FDA clearance.

PPG-Based Irregular Rhythm Notification (Software as Medical Device)

• Apple Watch: Irregular rhythm notification feature cleared as a software medical device. Operates passively using the optical sensor.
• Fitbit (Google): Irregular heart rhythm notification cleared by FDA.
• Samsung Galaxy Watch: Irregular heart rhythm notification cleared by FDA.

It is important to understand that FDA "clearance" (510(k)) is not the same as FDA "approval" (PMA). Clearance indicates that the device is substantially equivalent to a predicate device and is reasonably safe and effective. It does not indicate the same level of clinical evidence as full approval, which is typically reserved for high-risk devices. These wearable features are cleared for use as screening tools, not diagnostic devices. For how wearable data quality varies across devices, see our wearable comparison guides.

PPG-Only vs ECG: Sensitivity and Specificity Comparison

Understanding the performance differences between PPG-based screening and ECG-based classification is essential for interpreting what your wearable tells you.

PPG-Based Passive Screening

ECG-Based On-Demand Classification

The practical difference is that PPG screening catches episodes you did not know about (passive monitoring), while ECG provides higher-quality confirmation data when you suspect something is happening (active monitoring). The two features are complementary, and devices offering both (Apple Watch, Samsung Galaxy Watch) provide the most complete wearable AFib detection capability. Understanding the underlying PPG peak detection algorithms helps explain why sensitivity varies across conditions.

The False Positive Problem

False positives are the most significant practical challenge in wearable AFib screening. Even with high specificity (say 99%), screening a large population with low AFib prevalence generates many false positives. In a population of one million wearable users where 2% have AFib (20,000 people), a 99% specific algorithm would correctly identify 19,800 true positives but would also generate 9,800 false positive notifications among the 980,000 people without AFib.

Common causes of false positive AFib notifications include:

• Premature atrial contractions (PACs) and premature ventricular contractions (PVCs), which are extremely common and usually benign but create irregular pulse intervals
• Motion artifacts that corrupt the pulse interval measurements
• Sinus arrhythmia (normal respiratory variation in heart rate), particularly in younger, fit individuals
• Poor sensor contact during sleep or low-perfusion states

The psychological impact of false positives should not be underestimated. A notification suggesting AFib can cause significant anxiety, unnecessary emergency department visits, and downstream testing costs. This is why wearable algorithms are generally tuned toward higher specificity (fewer false positives) at the cost of lower sensitivity (more missed true cases). The clinical judgment is that missing some AFib episodes in an asymptomatic person is preferable to causing widespread alarm in healthy individuals.

Screening vs Diagnosis: Why the Distinction Matters

Wearable AFib detection is screening, not diagnosis. This distinction has important clinical and legal implications.

Screening means identifying individuals at elevated risk who should undergo further evaluation. A positive screen requires confirmatory testing before any treatment decisions. The goal is high sensitivity (catch as many true cases as possible) with acceptable specificity (not too many false alarms).

Diagnosis requires definitive evidence meeting established clinical criteria. For AFib, diagnosis requires documentation of the arrhythmia on a clinical-quality ECG (typically 12-lead) for at least 30 seconds, interpreted by a qualified clinician. A single-lead smartwatch ECG tracing can support a diagnosis but is not sufficient alone per most cardiology guidelines.

This distinction means that a smartwatch notification should never lead directly to treatment (anticoagulation, cardioversion) without proper clinical confirmation. The appropriate response to a positive wearable AFib screen is to consult a healthcare provider who can order confirmatory testing, evaluate stroke risk (CHA2DS2-VASc score), and make treatment decisions based on the complete clinical picture.

USPSTF and Clinical Guideline Perspectives

The U.S. Preventive Services Task Force (USPSTF) has evaluated screening for AFib with ECG in asymptomatic adults and concluded that the current evidence is insufficient to assess the balance of benefits and harms (I statement). This applies to both traditional ECG screening and wearable-based screening.

The key concern is not whether wearables can detect AFib (they can) but whether detecting AFib through screening in asymptomatic individuals leads to better health outcomes. The theoretical benefit is stroke prevention through earlier initiation of anticoagulation. The potential harms include bleeding complications from anticoagulation in people who might never have had a stroke, anxiety from false positives, and healthcare costs from downstream testing.

Several large randomized trials are underway to address this evidence gap, including the GUARD-AF trial and the Heartline study, which specifically evaluate whether smartwatch-based AFib screening reduces stroke rates compared to usual care. Until these results are available, clinical guidelines recommend that wearable AFib notifications be treated as opportunistic findings warranting clinical follow-up rather than as a basis for population-wide screening programs. For more context on HRV measurement methods and cardiac monitoring approaches, see our learning resources.

The Future of Wearable AFib Detection

Several technological and clinical advances are poised to improve wearable AFib detection over the coming years.

Algorithm improvements driven by larger training datasets and more sophisticated deep learning architectures will reduce false positive rates while maintaining or improving sensitivity. Transformer-based models processing raw PPG waveforms show promise for distinguishing AFib from PACs/PVCs more accurately than current interval-based approaches.

Continuous ECG monitoring through novel sensor designs (dry electrodes in watch bands, bioimpedance measurements) may eventually enable always-on ECG-quality monitoring without user interaction, bridging the gap between passive PPG screening and active ECG recording.

Integration with clinical workflows through standardized data formats (FHIR-compatible health records), direct physician notification systems, and AI-assisted triage will streamline the path from wearable detection to clinical confirmation. Some health systems are already piloting programs where smartwatch AFib detections are automatically routed to cardiology for remote review.

Stroke risk stratification using wearable data beyond rhythm detection, including AFib burden quantification (percentage of time in AFib), heart rate during AFib episodes, and activity context, will help clinicians make more nuanced treatment decisions. This is particularly important because AFib burden correlates with stroke risk, and a brief paroxysmal episode carries different implications than persistent AFib.

For those interested in related wearable health capabilities, our guide on cuffless blood pressure monitoring covers another frontier of cardiovascular wearable technology.

Frequently Asked Questions

How accurate is the Apple Watch at detecting AFib?

The Apple Watch irregular rhythm notification has a positive predictive value of approximately 84% based on the Apple Heart Study (Perez et al., 2019, NEJM). This means that when the watch flags a possible irregular rhythm, it is correct about 84% of the time. The on-device ECG feature (Apple Watch Series 4 and later) has been FDA-cleared with reported sensitivity of 98.3% and specificity of 99.6% for classifying sinus rhythm versus AFib in recordings of adequate quality. However, the ECG feature requires the user to actively take a 30-second recording and cannot detect AFib that occurs between recordings.

Can a smartwatch replace a Holter monitor for AFib detection?

No, a smartwatch cannot replace a Holter monitor for clinical AFib diagnosis. Holter monitors provide continuous, medical-grade ECG recording for 24-48 hours (or longer with extended monitors) with multiple lead configurations, detailed morphology analysis, and clinical-grade documentation. Smartwatch ECGs are single-lead (Lead I equivalent), 30 seconds long, and taken intermittently. However, smartwatches offer a significant advantage in long-term monitoring duration: a watch worn daily provides months to years of opportunistic screening, increasing the chance of catching paroxysmal AFib that a 24-48 hour Holter might miss.

Should I see a doctor if my smartwatch says I have AFib?

Yes, you should consult a healthcare provider for any persistent irregular rhythm notification from your smartwatch, but it is important not to panic. A single notification does not constitute a diagnosis. Your doctor will likely order a clinical ECG or extended monitoring (Holter or event monitor) to confirm or rule out AFib. Bring your smartwatch ECG recordings to the appointment, as they can provide useful context. False positives do occur, especially during exercise, after alcohol consumption, or with premature beats (PACs/PVCs) that are benign.

What is the difference between PPG-based and ECG-based AFib detection on a smartwatch?

PPG-based detection works passively in the background by analyzing pulse intervals from the optical heart rate sensor to identify irregular rhythms. It can screen continuously without user action but cannot provide a diagnostic-quality rhythm strip. ECG-based detection requires the user to actively touch electrodes and sit still for 30 seconds, producing a single-lead ECG tracing that can be shared with a physician and has higher diagnostic value. The two methods are complementary: PPG screens passively and flags potential episodes, while the ECG provides on-demand confirmation recordings.

Why does my smartwatch sometimes miss AFib episodes?

Smartwatches can miss AFib episodes for several reasons. PPG-based screening only analyzes data during periods of low motion, so AFib occurring during exercise or vigorous activity may not be assessed. Short paroxysmal AFib episodes lasting only minutes may fall between sampling windows. Motion artifacts can corrupt the PPG signal, causing the algorithm to discard potentially abnormal segments rather than risk a false positive. The ECG feature only captures 30-second snapshots when actively used, missing any AFib between recordings. Additionally, algorithms are generally tuned to minimize false positives, which inherently means some true episodes will be missed.