What a Scientific Study Says About Smartwatch aFib Accuracy

People often ask how accurate smartwatches are with this or that metric.

There are health tech reviewers who do such tests for a living, but they usually test watches on themselves only. This makes their results very individual and questionable – many individual parameters are involved, starting from skin tone, wrist size, subcutaneous fat, and many others. Your experience will most probably be different.

There is a known solution for this problem – a scientific approach. Multiple experiments, different participants (age, skin color, gender, health, lifestyle, and so on). Youtubers, even those who call themselves scientists, do not do that.

But even studies that follow a real scientific approach sometimes contradict each other.

To address this issue, scientists invented the “meta-analysis”. In meta-analyses, researchers collect as many relevant studies as they can find, filter out some with do not meet quality criteria, and then calculate statistics. If you search for reliable information, meta-analyses are your best choice.
There are not that many studies about smartwatches, and even fewer meta-analyses. Therefore, any of them is interesting, especially if the topic is not a “trivial” HR accuracy.

Recently, I found a meta-analysis about smartwatch atrial fibrillation (aFib) detection accuracy – that’s interesting!

Atrial fibrillation is an abnormal heart rhythm (arrhythmia) characterized by rapid and irregular beating of the atrial chambers of the heart. It often begins as short periods of abnormal beating, which become longer or continuous over time (Wikipedia).
If you didn’t know, aFib is a condition that the ECG in smartwatches can help to catch.

This meta-analysis is important news: unlike HR or GPS accuracy, regular reviewers cannot test aFib detection accuracy at all – none of them (at least none of whom I know) have aFib. Moreover, aFib symptoms are not permanent. Therefore, it is not that easy to test this smartwatch feature at all.

What are the results of this meta-analysis?

The short answer is: smartwatches are very reliable in aFib detection.

A bit longer answer: Smartwatches have a 95% correct positive rate and a 97% correct negative rate.
This means that:

If a smartwatch (any of them, including Garmins and Galaxy Watches I used most) detected aFib symptoms, the probability that this was correct is 95%.
If a smartwatch did not detect aFib symptoms, the probability that this was correct is 97%.

These are very good numbers! A note of caution: if you do not have aFib symptoms right now, it does not mean that they cannot show up in 10 minutes.

Other interesting information I found in this meta-analyses:

The Withings watches are not that great.
Different studies researched different models of the watch, usually already pretty outdated now.
The Apple Watch is the most researched watch model.
There is one study that claimed that it researched the “Samsung Galaxy Watch Active 3” – such a model did not exist at all – it was called the Samsung Galaxy Watch 3.
There is a smartwatch called “Seikon Epson Smartwatch” – never heard about it.
All these studies were done on people over 60 years old. How relevant these results are for younger persons is unclear.

If you want to read the analysis yourself (it is not a long paper), you can find it here: https://www.jacc.org/doi/10.1016/j.jacadv.2025.102133