Improving Diffusion Models for ECG Imputation with an Augmented Template Prior
This work addresses the issue of poor-quality ECG recordings in mobile health systems, which can degrade automated analysis, by providing a probabilistic imputation method that is incremental but offers practical gains.
The authors tackled the problem of imputing missing values in noisy ECG signals by developing PulseDiff, a template-guided diffusion model that incorporates subject-level templates and beat-level variations. Their method improved upon two diffusion baselines and matched or outperformed leading deterministic models on short-interval missing data in the PTBXL dataset.
Pulsative signals such as the electrocardiogram (ECG) are extensively collected as part of routine clinical care. However, noisy and poor-quality recordings are a major issue for signals collected using mobile health systems, decreasing the signal quality, leading to missing values, and affecting automated downstream tasks. Recent studies have explored the imputation of missing values in ECG with probabilistic time-series models. Nevertheless, in comparison with the deterministic models, their performance is still limited, as the variations across subjects and heart-beat relationships are not explicitly considered in the training objective. In this work, to improve the imputation and forecasting accuracy for ECG with probabilistic models, we present a template-guided denoising diffusion probabilistic model (DDPM), PulseDiff, which is conditioned on an informative prior for a range of health conditions. Specifically, 1) we first extract a subject-level pulsative template from the observed values to use as an informative prior of the missing values, which personalises the prior; 2) we then add beat-level stochastic shift terms to augment the prior, which considers variations in the position and amplitude of the prior at each beat; 3) we finally design a confidence score to consider the health condition of the subject, which ensures our prior is provided safely. Experiments with the PTBXL dataset reveal that PulseDiff improves the performance of two strong DDPM baseline models, CSDI and SSSD$^{S4}$, verifying that our method guides the generation of DDPMs while managing the uncertainty. When combined with SSSD$^{S4}$, PulseDiff outperforms the leading deterministic model for short-interval missing data and is comparable for long-interval data loss.