ECG criteria
- Assess whether the recording is single-lead — i.e., whether there is only a single waveform without chest leads
- Check device calibration data: compliance of 1 mV and recording speed with stated values (not always explicitly indicated as on a paper ECG; quantitative measurements are permissible only with known calibration and good signal quality)
- Assess rhythm: RR regularity, presence of P wave before each QRS
- Measure heart rate (HR) using RR intervals on the recording (the device usually calculates this automatically)
- Assess the device's algorithmic label (e.g., "sinus rhythm", "atrial fibrillation", "unreadable recording") as preliminary, not definitive
- Look for artifacts: baseline drift from tremor, movement, or poor electrode contact
What it looks like on ECG
A smartwatch recording with typical electrode placement (watch-wearing arm to opposite hand finger) resembles standard lead I, but the morphology and polarity of the P wave, QRS complex, and T wave may differ from textbook images — this depends on the arm the watch is worn on, arm positioning, contact quality, and the specific device. A strip lasting about 30 seconds is visible on the screen or in the app. Sampling frequency and amplitude calibration vary among different models and are not always explicitly stated, so a millimeter ruler from a paper ECG should not be directly applied to such a recording: quantitative measurements (mm, ms) are permissible only with known calibration and correct export of the recording; with low signal quality or unknown speed/gain parameters, such measurements are unreliable, and the recording should be assessed qualitatively — rhythm, rate, gross morphology.
Such a recording allows you to:
- assess rhythm regularity (regular/irregular);
- calculate heart rate using RR intervals;
- observe the presence or absence of individual P waves before each QRS (with good signal quality);
- notice gross abnormalities — pauses, marked tachycardia, overt AF with chaotic activity without P waves, and in some cases, noticeable ST-segment deviation in this single lead; however, such ST-T changes in a single lead are non-specific: they may be related to body position, electrode contact, artifacts, or an individual normal variant, and do not allow for independently diagnosing or ruling out ischemia.
Such a recording does not reliably allow you to:
- diagnose or rule out myocardial ischemia/infarction and determine its localization by walls (there are no leads over different heart walls); in the presence of symptoms, a standard 12-lead ECG and clinical evaluation are required regardless of what is visible in the single available lead;
- determine the electrical axis of the heart;
- diagnose hypertrophy and classify bundle branch blocks — one can only suspect a wide QRS or marked conduction disturbance requiring a 12-lead ECG;
- measure the QT interval with the same reliability as on a standard ECG: QT on a single-lead recording is measured less reliably due to motion artifacts, noise, and the inability to select the optimal lead or compare multiple leads, and should not be used for major clinical decisions without confirmation by a standard ECG (leads II or V5 are preferred with a high-quality 12-lead recording).
Mechanism
The watch and opposite-hand finger electrodes record the potential difference between two points of the body — a single resultant vector of cardiac depolarization and repolarization projected onto an axis close to that of lead I. With typical positioning, the recording direction is close to standard lead I, but this is not a full 12-lead ECG and not literally the same as Einthoven's lead I across all positions and devices: other limb and chest leads are missing, so it is not possible to reliably construct the frontal electrical axis (which requires at least two axes of Einthoven's triangle) or localize changes by heart walls. Consequently, most consumer watch algorithms are validated primarily for rhythm classification (sinus rhythm, atrial fibrillation, unreadable recording) and are not intended for diagnosing ischemia or systemic ST-T wall analysis; the raw single-lead waveform may contain information on ST-T changes, but recognition capabilities and validation depend on the specific model and software and must be verified separately.
Where it is used
- detecting episodes of paroxysmal atrial fibrillation in patients complaining of palpitations; wearable devices can be used as an auxiliary tool for such detection, especially in high-risk groups (including individuals over 65 with stroke risk factors), but they do not replace organized medical screening and subsequent physician confirmation;
- documenting an episode during a sensation of palpitations for subsequent review by a physician;
- dynamic self-monitoring in patients with known arrhythmias (e.g., post-AF ablation);
- in athletes — auxiliary recording of episodes during training (not a replacement for stress testing and Holter monitoring).
Differential diagnosis
- Motion/tremor artifact — a chaotic, irregular baseline where QRS complexes may be partially preserved or hidden by noise; a strong artifact is sometimes mistakenly classified by the algorithm as AF rather than flagged as an "unreadable recording", so an artifact cannot be distinguished from true AF based on a single automated label alone — the recording must be repeated at rest and the presence of regular QRS complexes on the raw waveform must be assessed;
- true atrial fibrillation — absence of P waves, irregular RR intervals, while QRS complexes remain discernible;
- sinus arrhythmia in young people and adolescents — RR varies smoothly and cyclically (often respiration-related) rather than chaotically as in AF;
- extrasystole captured in a short recording — a single premature complex is not grounds for an arrhythmia diagnosis without clinical evaluation and, if necessary, 24-hour monitoring.
Clinical significance and management
A smartwatch single-lead ECG is a screening tool, not a diagnostic one. A positive or borderline automated label (e.g., "atrial fibrillation") must be evaluated by a medical professional: the diagnosis is confirmed by a documented ECG, including a quality single-lead recording lasting at least 30 s reviewed by a specialist, or a 12-lead ECG, or ambulatory monitoring according to indications if the initial recording is unreliable or too short. Anticoagulant therapy should not be prescribed based solely on an automated device label.
When a patient presents with a smartwatch recording, the physician should:
- review the raw recording (not just the app's text label) and assess signal quality;
- if AF or another significant arrhythmia is suspected, order documentation with a standard ECG and/or Holter monitoring depending on the clinical situation;
- in the case of current or suspected chest pain, do not rely on a "normal" watch verdict: immediate emergency medical care must be sought, and a 12-lead ECG should be performed as soon as possible, repeated if necessary, and supplemented with troponin and other diagnostics — a single-lead recording does not rule out STEMI, occlusion equivalents, or NSTE-ACS;
- take the patient's context into account: in children and during pregnancy, algorithms and diagnostic thresholds may not be sufficiently validated, so the recording must be interpreted by a physician taking age or gestational age into account, and a standard ECG is required if pathology is suspected.
Hazards include false-negative results (missing a clinically significant arrhythmia or ischemia due to method limitations) and false-positive results (unwarranted anxiety and workups due to artifacts mistaken by the algorithm for arrhythmia).
Common interpretation errors
- taking the app's text label ("atrial fibrillation possible") as a definitive diagnosis without reviewing the waveform itself;
- measuring the QT interval or ST segment on a single-lead recording as if it were a standard ECG and using these figures for treatment decisions without confirmation by a standard recording;
- failing to account for motion and poor-contact artifacts that mimic an irregular rhythm or even simulate AF;
- ignoring current or suspected chest pain due to a "normal" smartwatch recording, delaying the search for emergency care;
- forgetting that the absence of chest leads precludes assessing the localization of changes by heart walls.
1. How many leads does a standard smartwatch ECG, such as the Apple Watch, record?
The recording is taken between the electrode on the watch and a finger of the opposite hand, yielding a single lead with an axis close to Einthoven lead I.
2. Which of the following cannot be reliably assessed using a single-channel smartwatch ECG?
The electrical axis of the heart is calculated from the ratios of multiple limb leads (at least two axes of Einthoven's triangle), which a single lead does not allow.
3. A patient presents a smartwatch recording with a "possible atrial fibrillation" notification and complains of chest pain. What is the correct physician management?
Chest pain requires emergency evaluation and a standard ECG to diagnose ischemia, while an arrhythmia notification from a single-channel device is merely a reason for confirmatory testing rather than a basis for immediate therapy.
Frequently asked questions
Can a smartwatch ECG diagnose a myocardial infarction?
No. A single-channel recording does not contain chest leads, so localization—and often the very presence—of ST-T ischemic changes cannot be reliably assessed from it, although gross changes are sometimes visible in the available lead. In the presence of ongoing or suspected chest pain, immediate medical care and a standard 12-lead ECG are required regardless of the smartwatch result.
How accurately do smartwatches detect atrial fibrillation?
In studies of selected users, validated models may show high sensitivity and specificity for detecting AF, but these metrics depend on the device, algorithm, and recording quality, and do not equal diagnostic accuracy in the general population. In tachycardia, extrasystoles, other arrhythmias, or artifacts, the algorithm may fail to classify AF. Any positive notification must be confirmed by an ECG reviewed by a medical professional before making treatment decisions.
How does a smartwatch ECG differ from Holter monitoring?
Holter monitoring continuously records one or more leads for hours or days and is better at detecting arrhythmias over time; ST-T assessment on such recordings is limited and depends on the specific system, while a standard 12-lead ECG is needed for ischemia diagnosis when indicated. Smartwatches provide a single lead and a short on-demand recording by the patient—this is a supplement, not a replacement for Holter monitoring.
Can a smartwatch ECG be used in a child or a pregnant woman?
Technically, recording is possible, but the algorithms and diagnostic thresholds of smartwatches may not be sufficiently validated for children and pregnant women; therefore, such a recording must be interpreted by a physician taking into account the patient's age or gestational age. If rhythm pathology is suspected, a standard ECG is required.
What to do if the smartwatch shows "poor reading"?
This is usually a motion, tremor, or poor electrode contact artifact rather than a sign of pathology. However, with severe artifacts, the algorithm sometimes erroneously gives other notifications, so pathology should not be ruled out based on a single automatic notification. The recording should be repeated at rest with a relaxed arm, and if repeated failures occur or symptoms persist, a standard ECG should be sought.
ECG Course
Interpretation from basics to myocardial infarction — learn to read any ECG using a systematic algorithm.
- Video lessons and theory for every topic
- Simulator case studies under every lesson
- Unlimited ECG simulator practice
- Study guide: "All ECGs in 25 minutes"
ECG Strip Simulator
ECG, radiology, histology, pathomorphology, pharmacology — thousands of flashcards with detailed explanations.
- Real ECG strips, not textbook drawings
- Instant answers and explanations
- Streaks and progress tracking in your dashboard
Related terms
More in the section "Children, Athletes, Pregnant Women"
Sources
- AHA/ACCF/HRS Recommendations for the Standardization and Interpretation of the Electrocardiogram, 2009
- ESC Guidelines for the diagnosis and management of atrial fibrillation, current edition
- Hampton J. The ECG Made Easy
Educational material for medical students and physicians. Does not replace clinical guidelines and physical examination; the ECG must always be interpreted in conjunction with the clinical picture.