AUTHOR=Li Li , Chen Xi , Hu Sanjun TITLE=Application of an end-to-end model with self-attention mechanism in cardiac disease prediction JOURNAL=Frontiers in Physiology VOLUME=Volume 14 - 2023 YEAR=2024 URL=https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2023.1308774 DOI=10.3389/fphys.2023.1308774 ISSN=1664-042X ABSTRACT=Heart disease has consistently been one of the major global health challenges, and early detection and prediction of heart disease are crucial for the survival and quality of life of patients. This research aims to develop an innovative method for predicting heart disease to meet the ongoing demand for accuracy and efficiency in healthcare. Our approach is based on end-toend deep learning, combining self-attention mechanisms and generative adversarial networks to enhance the predictive performance of heart disease. We have built an end-to-end model capable of taking various types of cardiac health data, including electrocardiograms, clinical data, and medical images. Self-attention mechanisms are incorporated into the model to capture correlations and dependencies among the data, thereby enhancing the understanding of latent features. At the same time, we use generative adversarial networks to synthesize additional cardiac health data to augment our training dataset. Our experiments are based on multiple publicly available heart disease datasets, which are used for model training, validation, and testing.During training, we utilize multiple evaluation metrics such as accuracy, recall, and F1-score to assess our model's performance. The results demonstrate that our model achieves accuracy rates exceeding traditional methods, surpassing 95% on multiple datasets. The recall metric shows that our model can better identify heart disease patients, with recall rates exceeding 90%.The F1-score, as a comprehensive evaluation metric, also exhibits outstanding performance, reaching optimal results. This research showcases the potential of end-to-end deep learning based on self-attention mechanisms in heart disease prediction. Our model excels across multiple datasets, offering new possibilities for improving early diagnosis and intervention in heart disease, thus enhancing patients' quality of life and health. This study holds broad clinical application prospects and promises substantial advancements in the healthcare field.1 Sample et al.