Avoiding Self-Diagnosis From Device Readings

Urgent first

Chest pain, severe breathlessness, fainting, or sudden face, arm or speech problems need emergency help, whatever your device says. In the US call 911 (or your local emergency number). See Symptoms That Need Assessment, Not Tracking. In the US, 988 reaches the Suicide & Crisis Lifeline. [1]

The short version

A consumer reading is a clue, not a conclusion. It can be wrong in both directions: an alert can be false, and a quiet device can miss a real problem. Labels used here: established, mixed, insufficient evidence, marketing claim, explained in How We Label Evidence in Biohacking.

Three kinds of number

What the evidence shows for one well-studied feature

Irregular-rhythm alerts are among the best studied consumer features, so they show both the promise and the limit. Evidence label: mixed, meaning accurate in some studies and still prone to false alerts.

• Apple Heart Study (2019). Among 419,297 participants, 0.52% received an irregular pulse notification. Of 450 notified people who returned a usable monitoring patch, atrial fibrillation was present in 34%. 57% of surveyed notified participants contacted a health care provider outside the study. [2] The study was funded by Apple. [2]
• Review of the Apple Watch ECG. A 2025 meta-analysis of 11 studies (4,241 participants, mean age about 63) reported pooled sensitivity of 94.8% and specificity of 95% against a standard 12-lead ECG. [3]
• Review across smartwatches. A 2025 meta-analysis of 26 studies (17,349 patients) reported overall sensitivity of 95% and specificity of 97%, with results varying by device. [4]

These figures belong to specific devices and study settings, and results varied widely between studies (statistical heterogeneity was high). [3][4] Do not carry them to other models or to people unlike those studied. Accuracy at detecting a rhythm is also not proof that wearing the device improves health. See Consumer ECG and Atrial Fibrillation Features.

Habits that reduce mistakes

• Treat one odd reading as a prompt to repeat the measurement later and to note how you feel, not as an answer.
• Look at patterns over time, not single nights or single beats. See Sleep Trends vs Individual Nights.
• Remember that every test can give false alarms. See False Positives, False Negatives and Test Accuracy and Multiple Testing and Overdiagnosis.
• Do not start, stop or change a medicine because of a reading. Talk with the prescriber.
• If a reading keeps you up at night, see Tracking Anxiety and Compulsive Measurement.