CardiacGen

CardiacGen is a Deep Learning framework for generating synthetic but physiologically plausible cardiac signals like ECG. Based on the physiology of cardiovascular system function, we propose a modular hierarchical generative model and impose explicit regularizing constraints for training each module using multi-objective loss functions. The model comprises 2 modules, an HRV module focused on producing realistic Heart-Rate-Variability characteristics and a Morphology module focused on generating realistic signal morphologies for different modalities. We empirically show that in addition to having realistic physiological features, the synthetic data from CardiacGen can be used for data augmentation to improve the performance of Deep Learning based classifiers.

This deep generative model comprises of two W-GAN’s, one inside each of HR2Rpeaks_Simulator and Rpeaks2EcgPpg_Simulator objects. Both of these are conditional generative models which incrementally add desired marginal information over the given conditional information.

How it works.

HR2Rpeaks_Simulator takes smooth Heart Rate (HR) averaged over a sliding window of 8s. and low-pass filtered along-with subject class as input condition and generates an R-peak train at Fs_out=100Hz with 1’s at R-peak locations and 0’s everywhere else. Internally, the W-GAN models uniformly-spaced tachograms at 5 Hz. Hence, HR2Rpeaks_Simulator adds subject-specific Heart Rate Variability (HRV) information to the input HR.

Rpeaks2EcgPpg_Simulator takes the R-peak train at Fs_in=100Hz. along-with subject class as input condition and generates an ECG signal at Fs_out=100Hz. Hence, Rpeaks2EcgPpg_Simulator adds subject-specific Morphological (Morph) information to the input R-peak train.

Currently, the W-GAN in HR2Rpeaks_Simulator has a single set of weights while the W-GAN in Rpeaks2EcgPpg_Simulator has subject specific fine-tuned weights.