Pragmatic Review on Progressions in Multimodal Disease Prediction with Combination of Machine Learning, Deep Learning and Electronic Health Records

Abstract

Machine Learning (ML) and Deep Learning (DL) methods can be integrated with multimodal data, particularly Electronic Health Records (EHR), presents a significant opportunity to enhance disease prediction. This review paper synthesizes key advancements, identifies gaps in multimodal disease prediction systems, and proposes a roadmap for integrating ML and DL models in clinical settings. By offering a critical comparison of current methods, this paper highlights challenges such as data integration, scalability, and real-world clinical application. In combination of data from various sources like Electronic Health Records (EHR), demographic and clinical dataset, many models have been reviewed and demonstrated with significant improvements in prediction performance compared to existing single-modality methods. This paper highlights the potential of multimodal disease prediction systems and discusses the research gaps that need to be addressed for broader clinical adoption and effectiveness.