Intelligent Clinical Decision-Making Systems have become a cornerstone of modern healthcare by enabling accurate diagnosis, prognosis and treatment planning through data-driven insights. With the growing availability of heterogeneous healthcare data such as medical images, clinical records, physiological signals and textual reports, multimodal learning has emerged as a powerful paradigm for integrating diverse data sources. This review presents a comprehensive and systematic analysis of multimodal approaches for intelligent clinical decision-making by leveraging machine learning, deep learning, transfer learning and natural language processing techniques. A structured literature search was conducted using IEEE, Elsevier, Wiley Online Library and Springer databases, focusing on peer-reviewed studies published between 2020 and 2025. The selected articles were analysed based on data modalities, learning strategies, healthcare applications, datasets and performance evaluation metrics. This review highlights the effectiveness of multimodal frameworks in addressing key challenges such as class imbalance, disease prediction, patient monitoring and treatment planning. Additionally, it discusses the benefits, open challenges and limitations of existing intelligent clinical decision frameworks, including scalability, interpretability and real-world deployment issues. Finally, the review outlines future research directions emphasizing the integration of Internet of Things-enabled healthcare data, federated learning for privacy preservation and blockchain-based secure data sharing to enhance the reliability and clinical adoption of intelligent decision-making systems.
Shehnaz et al. (Thu,) studied this question.