Arrhythmia Detection as a Digital Biomarker

Arrhythmia detection is the identification of an abnormal heart rhythm, most often atrial fibrillation, from sensors worn on the wrist or held against the body. It covers two related outputs: background screening, in which the device watches for irregular pulse patterns and notifies the wearer, and on demand recording, in which the wearer takes a single lead electrocardiogram that a clinician can later review.

This is the area where consumer digital measurement has moved closest to a regulated clinical role. Single lead electrocardiography features have cleared regulatory review in several markets, and irregular rhythm notification has been evaluated in prospective studies enrolling hundreds of thousands of participants, which is a scale almost no other digital measure in this library approaches.

Screening uses photoplethysmography. The device samples the pulse during quiet periods and applies an algorithm that looks for the beat to beat irregularity characteristic of atrial fibrillation. Because a false notification is disruptive, these algorithms are tuned conservatively and typically require repeated irregular readings before notifying.

Confirmation uses single lead electrocardiography. The wearer completes a circuit, usually by touching the device with the opposite hand, and the device records a short trace. That trace is a genuine electrocardiographic recording, though from one lead rather than twelve, which limits what can be concluded from it beyond rhythm.

Large studies have used the same design: notify on irregular pulse, then confirm with an ambulatory patch monitor, and report the proportion of notified participants in whom atrial fibrillation was confirmed.

Screening and case finding is the dominant application. Prospective studies including the Apple Heart Study, the Fitbit Heart Study and the Huawei Heart Study established that consumer devices can identify previously undiagnosed atrial fibrillation at population scale. Systematic screening programmes such as STROKESTOP have reported clinical outcomes from screening older populations.

The second application is post procedure follow up, where detection quantifies recurrence after ablation or cardioversion, and the third is symptom correlation, where a wearer records a trace during a palpitation episode to establish whether the symptom coincides with an arrhythmia. Detection is reported alongside symptom and functional instruments, because the correlation between detected events and how a person feels is weak.

A device does not diagnose. Confirmation requires clinical review, usually with a conventional electrocardiogram or a period of ambulatory monitoring, and acting on a notification alone is not appropriate. Single lead recordings show rhythm but cannot support conclusions that require a full twelve lead trace.

Detection is also selective. These systems target atrial fibrillation, and other arrhythmias may be missed or misclassified. Screening in low risk populations produces a substantial proportion of notifications that are not confirmed, generating anxiety and downstream investigation.

The unresolved issue is not accuracy. It is that population screening finds brief asymptomatic episodes faster than the evidence base can say what should be done about them.

• Perez MV, et al. Large-scale assessment of a smartwatch to identify atrial fibrillation. N Engl J Med. 2019. pubmed.ncbi.nlm.nih.gov
• Lubitz SA, et al. Detection of atrial fibrillation in a large population using wearable devices: the Fitbit Heart Study. Circulation. 2022. pubmed.ncbi.nlm.nih.gov
• Guo Y, et al. Mobile photoplethysmographic technology to detect atrial fibrillation. J Am Coll Cardiol. 2019. pubmed.ncbi.nlm.nih.gov
• Svennberg E, et al. Clinical outcomes in systematic screening for atrial fibrillation (STROKESTOP): a multicentre, parallel group, unmasked, randomised controlled trial. Lancet. 2021. pubmed.ncbi.nlm.nih.gov