JAMC: A jigsaw-based autoencoder with masked contrastive learning for cardiovascular disease diagnosis

Yue Ge, Huaicheng Zhang, Jiguang Shi, Deyu Luo, Sheng Chang, Jin He, Qijun Huang, Hao Wang

Abstract: Highlights•Instance discrimination-based contrastive learning is integrated with a jigsaw-based reconstruction task.•Contrastive and reconstruction tasks promote the fusion of global and local features.•Lead, temporal, and mixed jigsaw transformations are proposed and compared.•Lead mask transformation is introduced as a supplement for local lead correlation.•Morphological, temporal, and spatial physiological features of ECG signals are learned simultaneously.