부산대학교 도서관

This study investigates the automated identification of atrial fibrillation (AF) making use of electrocardiogram (ECG) data and convolutional neural networks (CNNs). The work achieves extraordinary sensitivity as well as specificity, which are crucial for differentiating AF from regular sinus beats, through thorough data preparation, curation, and CNN architecture optimization. These results offer great hope for early AF identification, which is essential for avoiding cardiovascular consequences. The requirement for extensive, diverse datasets and real-world clinical validation is highlighted by ongoing issues with data variety as well as model generalization. In order to ensure widespread use, the future path calls for the integration of wearable technology, interpretability in CNN models, along with regulatory compliance. In conclusion, this study represents a significant improvement in cardiovascular healthcare technology.