Artificial Intelligence–Driven Multimodal Modeling for Personalized Cardiac Therapy Optimization

Heart diseases have varying effects on people. The same treatment is not usually received equally by all patients. It is due to this that the standard treatment guidelines do not necessarily work well with everyone. Artificial intelligence (AI) has also been utilized in recent years to aid doctors in integrating various forms of health data, thereby enhancing patient care. This review describes how multimodal models based on AI can be the basis of individualized treatment in heart diseases. We consider typical sources of data applied in cardiac care, such as electrocardiograms, heart images, wearables, electronic health records, and omics data, such as genetics. We explain how AI techniques, in particular, deep learning and multimodal data fusion, integrate such data to learn more about each patient. Contrary to previous works, which primarily concentrate on the diagnosis, the current review points to the opportunity of AI to assist in the selection of the appropriate treatment, dosing, predicting the treatment outcome, and providing long-term care through continuous monitoring. Another topic that we address is explainable AI approaches that enable physicians to comprehend model choices and have confidence in AI-based systems. Besides, we address crucial issues, including discrepancies in the quality of data, risk-of-bias, absence of large clinical trials, and constraints of real-world application. We emphasize that safe and fair AI systems should be implemented that can be effective with various patient groups. Lastly, we map out the path of research in the future, including digital models of the heart, learning without privacy, and AI-directed clinical trials. On the whole, AI-based multimodal modeling can contribute to shifting heart care beyond general treatment principles to more adaptive, personalized, and effective therapy in clinical practice.