ECG Arrhythmia live demo

Select a real heartbeat below — each is drawn from a patient the model never saw during training. It is classified in your browser; nothing is uploaded.

Educational demonstration — not a diagnostic tool. This classifies pre-recorded research heartbeats to illustrate how a model reads an ECG. It cannot read a live ECG and is not a substitute for a clinician or an approved cardiac monitor. If you have symptoms or a heart-rhythm concern, seek medical care.

Select a heartbeat

Each beat is labeled with its true class (as annotated by cardiologists): N normal, S supraventricular, V ventricular, F fusion. Observe where the model succeeds — and where it fails.

What am I looking at?

Each trace is a single heartbeat — roughly a 0.7-second window of the ECG centered on the beat's sharp peak. The model was trained solely on the morphology of these windows. This is why it reads ventricular beats, which are visibly distinct, but struggles with supraventricular beats, whose shape is nearly normal and whose defining feature is timing — information the model cannot perceive. Addressing this is precisely the focus of the next iteration.