Artificial intelligence (AI)-based digital patient twins have the potential to make breakthroughs in research and clinical practices in rehabilitation. They make it possible to personalise treatment plans by simulating different rehabilitation scenarios and predicting patient-specific outcomes. DTs can continuously monitor a patient’s progress, adjusting therapy in real time to optimise recovery. They also facilitate remote rehabilitation by providing virtual models that therapists can use to guide patients without having to be physically present. Digital twins (DTs) can help identify potential complications or failures at an early stage, enabling proactive interventions. They also support the training of rehabilitation professionals by offering realistic simulations of different patient conditions. They can also increase patient engagement by visualising progress and potential future outcomes, motivating adherence to therapy. They enable the integration of multidisciplinary care, providing a common platform for different professionals to collaborate and improve rehabilitation strategies. The article aims to trace the current state of knowledge, research priorities, and research gaps in order to properly guide further research and shape decision support in rehabilitation. Keywords: rehabilitation; physiotherapy; digital twin; therapy planning; artificial intelligence; machine learning; health status prediction