Data Fusion Techniques for Multimodal Models in Medicine: A Review

Abstract

The increasing availability of diverse data sources in medicine necessitates sophisticated data fusion techniques for building comprehensive and accurate models. This review paper explores the landscape of data fusion techniques employed in multimodal medical models. These techniques are categorized based on the level of fusion: early, intermediate, and late. Prominent methodologies, including deep learning-based approaches, kernel methods, and probabilistic graphical models, are discussed along with their applications in medical image analysis, clinical decision support, and personalized medicine. Challenges associated with multimodal data fusion in medicine and potential future research directions are also outlined.