A Survey on Exploring Multimodal Fusion in Healthcare: Challenges and Solutions

Multimodal fusion in healthcare has emerged as a powerful tool for integrating diverse data sources, including medical images, sensor data, and clinical records, to enhance diagnostic accuracy, treatment planning, and patient monitoring. However, existing surveys need more discussion of the technical details of proposed multimodal fusion approaches and their practical applications. Therefore, we propose a comprehensive survey on using multimodal fusion in healthcare applications. It provides an in-depth examination of the current landscape, discussing key developments that have propelled the field forward as well as ongoing challenges that researchers and practitioners continue to overcome in their efforts to realize the full potential of this technology. The survey explores the fundamentals and core concepts underlying multimodal fusion, explores the methods and techniques that have been developed, and examines the broad applications in various healthcare domains. Moreover, it provides a detailed discussion of the multimodal deep learning classification pipeline, highlighting the critical steps involved in pre-processing, feature extraction, information fusion, and model evaluation. Furthermore, the survey critically analyzes the data integration challenges that remain significant obstacles and identifies promising future directions that hold the potential to shape the future of this rapidly evolving field.