Agentic AI-driven human digital twins for personalized healthcare: a comprehensive survey

Abstract

Abstract Human Digital Twins (HDTs) are an innovative concept in the field of personalized medicine, relying on real-time models of a human body formed via collection and analysis of physiological, clinical, behavioural, and environmental data. Research on HDTs gives evidence to a considerable promise in terms of disease predictions, patient management, therapy optimization, and clinical decision-making. Nevertheless, existing technologies suffer from incomplete multimodal data integration, low-level of independent intelligence, missing reasoning possibilities, and non-adaptive decision making, which are vital for the development of sophisticated smart healthcare systems. Technological advancements are bringing in a new incorporating idea of Agentic Artificial Intelligence (Agentic AI) and Multimodal Foundation Models (MFMs), which may help to solve the outlined limitations and to bring autonomous reasoning, multimodal learning, and intelligent support to clinical practice. This survey aims to review Agentic AI-led HDTs for personalized medicine, concentrating on the interrelation between HDTs and Multimodal Foundation Models, Large Language Models, Vision-Language Models, and autonomous artificial intelligence agents Our study discusses the history, models, supporting technologies, multiple data sources, and clinical uses of HDTs, along with the analysis of Agentic AI approaches that can be applied for flexible decision-making, continuous patient monitoring, and individualized treatment development. Besides, this survey consists of comparative analysis of existing research works where methods, advantages, disadvantages, and research challenges are specified. There are also serious issues discussed, like interoperability, justification, confidentiality ensuring, safety, ethical regulation, clinical testing and legal compliance, etc. And finally, we specify prospects of research within the fields of self-adaptive digital twins, reliable autonomous health agents, federated intelligence, edge AI and multimodal foundations focused on healthcare.