Synthetic ECG signals generation: A scoping review

Beatrice Zanchi, Giuliana Monachino, Luigi Fiorillo, Giulio Conte, Angelo Auricchio, Athina Tzovara, Francesca D. Faraci

Abstract: Highlights•ECG synthesis uses three approaches: mathematical models, computer vision, or deep generative models.•Mathematical models offer high signal fidelity but limited variability in ECG generation.•Computer vision inherited methods excel in ECG data augmentation due to simple implementation.•Deep generative models lead ECG synthesis, offering both high fidelity and signal variety.•No standard metrics exist to comprehensively assess the quality of synthetic ECGs.