The field of cardiovascular disease research is moving towards the development of more accurate and reliable diagnostic tools, leveraging advances in artificial intelligence and machine learning. Recent innovations have focused on improving the analysis of medical imaging and physiological signals, enabling automated detection and diagnosis of cardiovascular conditions. A key trend is the integration of multimodal data and adaptive algorithms to refine personalized care. Notably, hybrid models combining deep learning architectures with traditional machine learning techniques have shown promising results. In particular, the use of transformer layer fusion and attention-based mechanisms has improved the prediction of ICU in-hospital mortality. Furthermore, the application of gradient boosting models and convolutional neural networks has enhanced the detection of atrial fibrillation. Noteworthy papers include:

• DeepBoost-AF, which introduces a novel hybrid methodology for robust atrial fibrillation detection.
• Predicting ICU In-Hospital Mortality Using Adaptive Transformer Layer Fusion, which presents a modular attention-based architecture for early identification of high-risk ICU patients.