ECG Inpainting with denoising diffusion prior

Lisa Bedin, Gabriel Cardoso, Remi Dubois, Eric Moulines

Keywords: diffusion models, electrocardiogram, generative models, monte carlo

Abstract: In this work, we train a generative denoising diffusion model (DDGM) in healthy electrocardiogram (ECG) data capable of generating realistic healthy heartbeats. We then show how recent advances in solving linear inverse Bayesian problems with DDGM can be used to derive interpretable outlier detection tools for electrophysiological anomalies.

Submission Number: 21