ProEGAN-MS: A Progressive Growing Generative Adversarial Networks for Electrocardiogram Generation

Electrocardiogram (ECG) is a physiological signal widely used in monitoring heart health, which of great significance to the detection and diagnosis diseases. Because abnormal rhythms are very rare, most ECG datasets have data imbalance problems. At present, many algorithms for anomaly automatic recognition affected by imbalance. Conventional augmentation methods not suitable signal, because one-dimensional their morphology has significances. In this paper, we propose ProGAN based sample generation model, called ProEGAN-MS, solve problem The model can stably generate realistic samples. We evaluate fidelity diversity generated compare distribution original data. addition, order show more intuitively, manually checked calculate statistics results that compared with other on GANs, our higher diversity, samples closer Finally, established neural network models arrhythmia classification, them improvement classification performance ProEGAN-MS. augmented ProEGAN-MS effectively improve insufficient sensitivity precision model.