This study aims to develop novel causal models for common deep learning tasks in electrocardiogram (ECG) analysis. The motivation stems from the understanding that these tasks are interdependent and ,by integrating well-established medical knowledge into deep learning methodologies, this approach seeks to bridge the gap between traditional medical practices and advanced computational techniques. Specifically, two causal maps are proposed to guide the design and implementation of problem-solving strategies. The performance of the resulting deep learning models is then evaluated and compared in testing scenarios.