Interpretable Multimodal AI Framework for Predicting Chronic Disease Progression

Chronic disease prediction has gained significant momentum with the rapid advancement of Internet of Medical Things (IoMT) and Healthcare 5.0, enabling continuous patient monitoring and large-scale health data acquisition. The integration of deep learning and machine learning models has shown strong potential in early disease diagnosis, progression analysis, and personalized treatment planning. While these intelligent systems improve predictive accuracy and clinical efficiency, their increasing complexity introduces challenges related to model interpretability, data heterogeneity, scalability, and clinical trust. This survey presents a structured and analytical review of chronic disease prediction methodologies in IoMT-enabled healthcare systems, with primary emphasis on multimodal data fusion, deep neural architectures, temporal health data modeling, and explainable AI frameworks. Mathematical formulations are discussed to provide theoretical grounding for feature extraction, temporal dependency learning, and disease risk estimation models. Furthermore, commonly used healthcare datasets, evaluation metrics, system architectures, open challenges, and emerging research directions are critically examined from a Healthcare 5.0 and clinical decision-support perspective. This study aims to serve as a comprehensive reference for researchers and practitioners working on trustworthy, interpretable, and scalable AI-driven healthcare systems, with Alzheimer’s disease considered as a representative chronic disease case study.