Overview
ECGs are synthesized voltage functions, not recordings from patients. A base rhythm supplies event timing; lead-specific morphology supplies P/QRS/T shape, axis, width, amplitude and repolarization patterns. Disease-associated patterns are illustrative possibilities rather than diagnostic certainty.
The interpretation is curated from the selected rhythm, finding and variant. It does not run an ECG diagnostic algorithm on pixels, infer coronary anatomy from a pressure trace or consult a GPT model at runtime.
Voltage construction
Base complexes use timed analytic voltage components. Sinus atrial timing is handled separately so enlarged/notched P waves retain a plausible PR relationship. Sequential atrial/ventricular pacing adds separate stimuli and captured complexes. NSVT replaces the baseline complex only during the scheduled run.
Lead-specific findings and variant profiles add representative bundle-branch block, chamber loading, scar, ST/T abnormalities or VT morphology. Telemetry II and V1 use the same finding family as the twelve-lead display; they are not simply identical traces with different labels.
Patterns and variants
Compatible variants are filtered by rhythm. RVOT, LVOT and fascicular VT examples alter axis and precordial morphology; these patterns illustrate localization clues without modeling a conduction network or proving a clinical origin. ST/T variants do not automatically change coronary perfusion, troponin or wall-motion territory.
The paused final read, finding highlights and segment tooltips are generated from metadata associated with the visible example. Rhythm-wide interpretation and a local P/PR/QRS/ST/T explanation answer different questions.
Sampling and calibration
ECG sampling uses an absolute 500 Hz time lattice, with extra samples around pacing spikes. This prevents the scrolling window from moving the sample grid through a QRS peak. Bounded caches are invalidated when the clinical model changes.
Sweep and gain change display geometry, not conduction intervals or the underlying voltage. A teaching gain makes small features visible but should be read with its displayed calibration. Extreme rates can compress normal phase relationships; morphology is idealized rather than a biophysical activation solution.
References & further reading