Methods and limitations

What HeartPro calculates, what is unvalidated, and how to interpret source data.

Source data and original waveform

HealthKit supplies the original single-lead ECG voltage series, sample timing and Apple's recorded classification. HeartPro displays that source classification; it does not perform Apple's atrial-fibrillation classification. Waveform views rescale the supplied data for display and support manual two-point measurements.

Automatic estimates

HeartPro runs deterministic Swift signal processing locally. It checks signal quality, estimates R peaks and RR intervals, compares timing and local waveform shape, and estimates eligible P, PR, QRS, ST/J, T, QT, QTc and other intervals. Boundary ambiguity, insufficient duration, noise and overlapping waves can suppress an estimate. Missing values are not replaced with zero. U-wave candidates appear only when enough repeatable, separated candidates exist.

Candidate premature beats combine timing and waveform rules. Wide/narrow categories do not establish ventricular/atrial origin. The rhythm reference score is the proportion of beats not flagged by these rules; it is not a health score, risk estimate or assurance that a rhythm is safe. Both false positives and missed events can occur.

Reference ranges are general information. Neither the interface nor exported PDFs infer disease causes or offer medication or treatment explanations from an individual measurement. Deviations are not presented as personal diagnoses. A usage notice appears before first use and remains available in Settings. This notice does not replace algorithm validation or applicable regulatory requirements.

Two different HRV sources

Apple Health HRV uses the SDNN samples supplied by HealthKit. HRV estimated from a single short ECG is separate: it uses eligible intervals and does not bridge rejected gaps for adjacent-interval statistics. The filtering rules cannot confirm normal beats. A roughly 30-second estimate is not equivalent to standardized five-minute HRV or its reference values.

Evidence status

These automatic calculations have not been validated against independently annotated real Apple Watch recordings. A prior development comparison with eight public MIT-BIH recordings reported 43.78% overall premature-beat sensitivity and substantial missed events in complex rhythms. Those Holter recordings differ from Watch single-lead ECGs; this is a limited development benchmark, not a clinical accuracy claim or representative Watch validation.

No regulatory clearance for HeartPro is claimed. Retaining these features does not establish readiness for App Store review. Watch accuracy evidence is a release prerequisite. Numerical agreement on synthetic signals, UI tests and PDFs do not establish medical performance.

Making decisions

HeartPro cannot reliably identify or exclude a heart attack, ischemia or other heart disease. A high score, no marked candidates, missing measurement or Apple's sinus classification does not mean you are safe. Consult a doctor before medical decisions. For persistent chest pain, severe breathlessness or fainting, seek emergency help immediately.

References

These references explain source data and concepts. They do not validate HeartPro's implementation.