PPG-CAGAN: A PPG signal with coronary artery disease synthesis method based on generative adversarial learning

Coronary artery disease (CAD) is a leading cause of morbidity and mortality worldwide. Photoplethysmography (PPG) has become an important tool for cardiovascular disease monitoring due to its ease of measurement. However, studies on the CAD detection based on PPG signals face challenges such as data scarcity and class imbalance, which limit the accuracy and robustness of models. In this study, we propose a novel generative adversarial network called PPG-CAGAN for generating PPG signals. The model incorporates Bi-LSTM, Res-CBAM, and Parallel Multi-Scale Attention mechanism (PMSA) to improve the accuracy of PPG signal generation. Experimental results show that PPG-CAGAN achieves excellent performance on multiple evaluation metrics for generating PPG signals of different lengths. Compared to other data augmentation methods, PPG-CAGAN generates more realistic PPG signals, effectively enhancing the generalization ability of coronary artery disease detection models. Therefore, the data augmentation method proposed in this study effectively addresses the issues of PPG data scarcity and class imbalance for coronary artery disease, which is of great significance for improving the performance of the intelligent detection model of CAD based on PPG. • Novel PPG-CAGAN model for generating high-quality PPG signals. • Hybrid Bi-LSTM, Res-CBAM, and PMSA modules enhance the performance of the PPG generative model. • Effectively solves data scarcity and class imbalance in CAD detection.