Early Prediction of Chronic Diseases Through Multimodal Data Fusion in Machine Learning: A Survey

Chronic diseases like diabetes and cardiovascular diseases develop slowly and may go untreated during the later stages. This paper aims at providing a wide overview of the application of Multimodal Data Fusion (MDF) in machine learning (ML) for the early onset of chronic diseases. MDF incorporates multiple type of health data sources i.e. EHRs, wearables, genomics, and medical imaging to increase the accuracy of the prediction and thus the timing of interventions. The role of deep learning, explainable AI, and federated learning in enhancing disease detection is explored, while also addressing challenges such as data heterogeneity, lack of interpretability in models, and privacy risks. Moreover, the study reveals the advantages of the combination of AI, edge computing, and real-time health care monitoring as the future of better, accurate and individualized healthcare. The remaining gaps are identified and the directions for future work include standardization, real-world implementation, and ethical issues. This survey can be used as a reference for the development of AI-based early diagnosis and prevention, which can help to enhance the health of populations.