What is the most accurate wearable for AFib detection?
Samsung and Amazfit share first place for AFib detection, with Apple third. The Amazfit result needs context: it is based on older ECG-capable health devices sold mainly in China, not the current global Amazfit lineup. The leading brands are close, and the largest meta-analysis found that brand differences were no longer statistically significant after adjusting for sample size.
AFib detection accuracy by brand
This ranking measures how accurately wearable algorithms identify atrial fibrillation against clinical ECG reference diagnoses. It covers AF detection rather than general ECG waveform quality, and studies may use single-lead ECG, PPG or both depending on the device.
Rank | Brand | Verdict |
|---|---|---|
1 | Top-tier pooled diagnostic accuracy, with the highest SROC AUC in the 2025 meta-analysis | |
1 | Top-tier pooled result from older China-focused ECG devices; current global Amazfit watches do not offer ECG | |
3 | Very strong accuracy plus the broadest independent validation base among mainstream watches | |
4 | Good overall specificity, but pooled sensitivity trails the leading brands | |
5 | Can classify AF well when a tracing is conclusive, but inconclusive readings materially reduce real-world diagnostic yield | |
NR | Promising evidence exists, but it is too narrow to generalize across current Garmin ECG-capable watches | |
NR | Insufficient connected brand-specific criterion-validation evidence for a separate Pixel Watch position |
How the ranking was decided
AFib detection is a diagnostic-classification problem, so both sensitivity and specificity matter. Sensitivity tells us how often a wearable correctly identifies AF, while specificity tells us how often it correctly recognizes a non-AF rhythm. We also treat inconclusive readings as important because a watch that frequently refuses to classify a tracing is less useful in practice even if its accuracy looks excellent after those tracings are removed.
The main anchor is the 2025 systematic review and diagnostic meta-analysis, which combined 26 studies and 17,349 patients. Samsung had the highest reported brand-level SROC AUC at 0.983, while Amazfit was essentially alongside it at about 0.98. The same analysis reported pooled sensitivity and specificity of 96.59% and 95.81% for Samsung, and 98.68% and 98.87% for Amazfit.
Samsung and Amazfit therefore share first rather than being separated by a cosmetic decimal-point difference. That is especially important because the meta-analysis found that apparent differences between brands were no longer statistically significant after adjustment for sample size. However, the Amazfit subgroup represents older ECG-capable Health Band and Health Watch-era devices sold mainly in China. Current global Amazfit watches do not offer ECG, so this rank should not be used to infer AFib capability or accuracy for today's Amazfit range.
Apple ranks third, but the gap is small. The meta-analysis reported 93.62% sensitivity, 96.97% specificity and an AUC of 0.968 for Apple Watch. Apple also performs well in direct clinical comparisons. In the BASEL Wearable Study, Apple Watch Series 6 produced 85% sensitivity and 75% specificity when inconclusive results were counted in the intention-to-diagnose analysis, and its performance rose substantially when only conclusive tracings were assessed.
Withings ranks fourth. Its pooled specificity is strong, but pooled sensitivity of 88.58% is lower than the leading brands. The three-smartwatch comparative study and BASEL comparison both support the same broad picture: Withings can identify AF usefully, but its automatic classifier is more likely to miss cases than the leaders.
Fitbit ranks fifth mainly because of inconclusive tracings. In the BASEL study, Fitbit Sense showed 66% sensitivity and 79% specificity when every attempted recording counted, but among conclusive tracings its sensitivity and specificity were much stronger. That difference is exactly why the ranking does not simply cherry-pick the best-looking number from each paper.
Garmin remains unranked despite a promising result in the meta-analysis because the connected evidence is too narrow to generalize across Garmin as a brand, particularly to newer ECG-capable watches. Google also remains unranked because the available database does not yet contain enough Pixel Watch-specific AF validation. Newer clinical work such as the 2025 analysis of inconclusive smartwatch ECGs also reinforces a broader limitation: baseline ECG abnormalities can materially increase the chance that consumer devices return an inconclusive automated result.
Supporting studies
SamsungSamsung Galaxy Watch family, Samsung Galaxy Watch 3, Samsung Galaxy Watch Active 34 studies
Samsung shares first place. The 2025 meta-analysis reported pooled sensitivity of 96.59%, specificity of 95.81% and the highest brand-level SROC AUC at 0.983. Direct comparative studies also place Galaxy Watch devices among the stronger automated AF classifiers.
- Meta-analysis November 1, 2025Accuracy of Smartwatches in the Detection of Atrial Fibrillation: A Systematic Review and Diagnostic Meta-Analysis
Pooled AF-detection sensitivity was 96.59% and specificity 95.81% for this device/brand subgroup.
- Clinical ECG April 29, 2023Clinical validation of an artificial intelligence algorithm offering cross-platform detection of atrial fibrillation using smart device electrocardiograms
Manufacturer algorithm produced a conclusive classification in about 85% of tracings; the cross-platform AI algorithm increased the conclusive rate substantially.
- Clinical ECG January 18, 2023Clinical Validation of 5 Direct-to-Consumer Wearable Smart Devices to Detect Atrial Fibrillation: BASEL Wearable Study
Galaxy Watch 3 automated AF detection showed 85% sensitivity and 75% specificity.
- Clinical ECG February 1, 2022Smartwatch Electrocardiograms for Automated and Manual Diagnosis of Atrial Fibrillation: A Comparative Analysis of Three Models
Samsung automated AF classification showed 88% sensitivity and 81% specificity.
AmazfitAmazfit smartwatches1 study
Amazfit shares first place on the available pooled evidence, with 98.68% sensitivity, 98.87% specificity and an AUC of about 0.98 in the 2025 meta-analysis. This needs an important qualification: the Amazfit evidence comes from older ECG-capable health devices, including the Health Band 1S and Health Watch-era products sold mainly in China. It should not be read as evidence that the current global Amazfit lineup offers ECG or AFib detection.
- Meta-analysis November 1, 2025Accuracy of Smartwatches in the Detection of Atrial Fibrillation: A Systematic Review and Diagnostic Meta-Analysis
Pooled AF-detection sensitivity was 98.68% and specificity 98.87% for this device/brand subgroup.
AppleApple Watch family, Apple Watch Series 6, Apple Watch Series 55 studies
Apple ranks third. Its pooled sensitivity was 93.62% and specificity 96.97%, with an SROC AUC of 0.968. Apple also has the deepest collection of independent smartwatch AF studies, including direct comparisons against 12-lead ECG.
- Meta-analysis November 1, 2025Accuracy of Smartwatches in the Detection of Atrial Fibrillation: A Systematic Review and Diagnostic Meta-Analysis
Pooled AF-detection sensitivity was 93.62% and specificity 96.97% for this device/brand subgroup.
- Clinical ECG January 1, 2025Comparative Evaluation of Consumer Wearable Devices for Atrial Fibrillation Detection: Validation Study
Apple Watch detected AF with 100% sensitivity and 97.8% specificity in this study.
- Clinical ECG April 29, 2023Clinical validation of an artificial intelligence algorithm offering cross-platform detection of atrial fibrillation using smart device electrocardiograms
Manufacturer algorithm produced a conclusive classification in about 84% of tracings; the cross-platform AI algorithm increased the conclusive rate substantially.
- Clinical ECG January 18, 2023Clinical Validation of 5 Direct-to-Consumer Wearable Smart Devices to Detect Atrial Fibrillation: BASEL Wearable Study
Apple Watch Series 6 automated AF detection showed 85% sensitivity and 75% specificity.
- Clinical ECG February 1, 2022Smartwatch Electrocardiograms for Automated and Manual Diagnosis of Atrial Fibrillation: A Comparative Analysis of Three Models
Apple Watch automated AF classification showed 87% sensitivity and 86% specificity.
WithingsWithings smartwatches, Withings ScanWatch, Withings Move ECG4 studies
Withings ranks fourth. The pooled meta-analysis showed 88.58% sensitivity and 95.03% specificity for the Withings subgroup. Direct studies confirm useful AF detection, but sensitivity is less consistent than it is for the top three brands.
- Meta-analysis November 1, 2025Accuracy of Smartwatches in the Detection of Atrial Fibrillation: A Systematic Review and Diagnostic Meta-Analysis
Pooled AF-detection sensitivity was 88.58% and specificity 95.03% for this device/brand subgroup.
- Clinical ECG April 29, 2023Clinical validation of an artificial intelligence algorithm offering cross-platform detection of atrial fibrillation using smart device electrocardiograms
Manufacturer algorithm produced a conclusive classification in about 74% of tracings; the cross-platform AI algorithm increased the conclusive rate substantially.
- Clinical ECG January 18, 2023Clinical Validation of 5 Direct-to-Consumer Wearable Smart Devices to Detect Atrial Fibrillation: BASEL Wearable Study
Withings ScanWatch automated AF detection showed 58% sensitivity and 75% specificity.
- Clinical ECG February 1, 2022Smartwatch Electrocardiograms for Automated and Manual Diagnosis of Atrial Fibrillation: A Comparative Analysis of Three Models
Withings automated AF classification showed 78% sensitivity and 80% specificity.
FitbitFitbit ECG devices, Fitbit Sense3 studies
Fitbit ranks fifth because inconclusive classifications matter. In the BASEL Wearable Study, Fitbit Sense reached much stronger sensitivity and specificity when inconclusive tracings were excluded, but intention-to-diagnose performance fell to 66% sensitivity and 79% specificity when those unresolved readings counted against the device.
- Meta-analysis November 1, 2025Accuracy of Smartwatches in the Detection of Atrial Fibrillation: A Systematic Review and Diagnostic Meta-Analysis
Pooled AF-detection sensitivity was 65.57% and specificity 79.29% for this device/brand subgroup.
- Clinical ECG April 29, 2023Clinical validation of an artificial intelligence algorithm offering cross-platform detection of atrial fibrillation using smart device electrocardiograms
Manufacturer algorithm produced a conclusive classification in about 81% of tracings; the cross-platform AI algorithm increased the conclusive rate substantially.
- Clinical ECG January 18, 2023Clinical Validation of 5 Direct-to-Consumer Wearable Smart Devices to Detect Atrial Fibrillation: BASEL Wearable Study
Fitbit Sense automated AF detection showed 66% sensitivity and 79% specificity.
GarminGarmin Forerunner 9451 study
Garmin remains unranked. The meta-analysis includes a strong sensitivity result for a Forerunner 945 subgroup, but the connected evidence is too narrow and does not provide enough replication across Garmin generations or current ECG-enabled models for a defensible brand-wide position.
- Meta-analysis November 1, 2025Accuracy of Smartwatches in the Detection of Atrial Fibrillation: A Systematic Review and Diagnostic Meta-Analysis
Pooled AF-detection sensitivity was 96.97% and specificity 88.24% for this device/brand subgroup.