What are the benefits and challenges of using artificial intelligence (AI) in neurorehabilitation? A very rapid review of the literature

Abstract

The use of artificial intelligence (AI) is growing across disciplines and becoming increasingly discussed in neurorehabilitation. To capture the latest developments in order to understand which, if any, solutions are sufficiently developed for use in practice, we conducted a very rapid literature review, systematically searching the Embase and MEDLINE databases. The five publications that met the criteria for review point to most recent developments in improving diagnosis and prognostication using AI, with no studies examining AI-based rehabilitation interventions directly. However, there was a theoretical ambition of ingraining this technology in rehabilitation programmes themselves in the future. AI has demonstrated superior predictive power compared to traditional approaches when built on large subsets of patient outcome data and was revealed beneficial in estimating the location and extent of brain damage using brain scans. Nevertheless, the quality of the current evidence is limited by lack of follow-up studies of and lack of variability within the study samples, which reduces generalisation to certain groups, such as those with complex needs.