Analysis of Multi-modal Data Through Deep Learning Techniques to Diagnose CVDs: A Review

Abstract

In cardiology, there has been a surge in artificial intelligence (AI), machine learning, and deep learning techniques. Artificial intelligence (AI) and electronic health records have the potential to advance our knowledge of disease states and enable personalized cardiac care in the era of modern medicine. With its latest data fusion techniques of non-imaging and imaging data (including cardiac magnetic resonance imaging, echocardiography, and cardiac computed tomography), the field of cardiac medicine is evolving, leading the revolution in precision cardiology. Although these data were previously used in isolation, new developments in deep learning (DL) and machine learning (ML) allow these data sources to be integrated to generate multimodal insights. There is growing interest in the application of data fusion, which uses ML and DL techniques to integrate data from multiple modalities into cardiac care.  We review the most advanced research in this paper, emphasizing how the new methods of data fusion are delivering clinical and scientific insights uniquely to the field of cardiovascular medicine. Although multi-modal deep learning yields more reliable estimations than multi-modal machine learning and unimodal techniques, it suffers from limitations related to scalability and the time-consuming nature of concatenating information.