Current value of Artificial Intelligence in Physical and Rehabilitation Medicine: A systematized narrative review

Background
Functional limitations and mobility impairments arising from a wide range of conditions represent a growing public health challenge and contribute to an increased demand for Physical and Rehabilitation Medicine (PRM). This demand aligns with the framework of the World Health Organization’s 'Rehabilitation 2030' initiative, which emphasizes equitable access to rehabilitation as a core global health priority. Artificial intelligence (AI) could offer promising opportunities to support this objective. In this context, exploring how AI is currently implemented in PRM is essential to better understand its documented contributions to date and future potential.
Objectives
The main objectives of this systematized narrative review are to (1) synthetize current studies exploring how AI is used in PRM, and (2) assess its relevance by identifying existing knowledge gaps to inform future research directions and potential applications in the field.
Material and Methods
Medline (Ovid) and Embase (www.embase.com) were searched in order to find studies published from 2020, focused on AI and PRM. Published studies in French or in English were included.
Results
A total of 87 references were found, including 32 on Medline and 55 on Embase. After removing 19 duplicates, 68 references were screened based on title and abstract. Full-text assessment was performed for 53 articles, leading to the inclusion of 37 studies. Currently, only a few articles show that, although some AI tools are already used in limited contexts in PRM, their integration remains mainly exploratory and potential.
Conclusions
To date, most articles in which AI is discussed in PRM report potential leads, without practical use or reported prospective clinical validation. A hybrid approach, combining human judgment and empathy with AI’s efficiency, could substantially contribute to improving standardization and clinical decision support in PRM settings as well as improving education and research in rehabilitation.
Conflict of interest
The authors declare no conflict of interest.