Artificial intelligence in upper-limb, language and cognitive neurorehabilitation after stroke

PURPOSE OF REVIEW: Artificial intelligence (AI) is being increasingly integrated into upper-limb, language and cognitive neurorehabilitation research and there is interest in where the research currently lies along the pipeline towards clinical implementation and real-world impact. RECENT FINDINGS: Across these three domains of neurorehabilitation, the research into the use of AI has shown clear potential to enhance assessment, personalize rehabilitation and extend access to care but is yet to demonstrate clinical utility and superiority over other digital rehabilitation approaches. There remain concerns about small and selective datasets being used for modelling, privacy and equity issues and sustainable implementation. It is suggested that AI should be positioned as an important tool in the clinical infrastructure rather than an autonomous clinical decision maker within clinical practice. SUMMARY: Future research across these neurorehabilitation domains should consider implementation risks and needs early in the research pipeline, by using ethical principles and guidelines to ensure transparency and consumer involvement in AI research and implementation into clinical practice.