Artificial intelligence in rehabilitation: a living systematic mapping review - first release.
Giovanni Morone, Riccardo Carbonetti, Alex Martino Cinnera and 3 others
PMID 41424220WHAT IT FOUND
Most AI research in rehabilitation is small, internal, and focused on stroke or gait, with very few trials testing real clinical benefit.
Half the studies had no comparison group, and external validation was rare, so these tools are not yet proven to improve patient outcomes.
Key findings
01Only 10.4% of included studies used formal reporting tools, and just 5.8% performed external validation, meaning most AI models have not been tested on new, independent patient groups.
02Half of the studies (50.8%) did not include any comparison group, making it impossible to tell if AI tools performed better than standard care or other algorithms.
03Neurological rehabilitation, especially stroke, dominated the field (29.6% of studies), while usability and clinical outcome measures were reported in only 5.7% of studies.
STILL TO COME
How it was doneWhat they foundWhat it means for PTsWhat it means for OTsWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The review is a mapping study, not an effectiveness study. It describes what has been done, not what works. No meta-analysis or effect-size synthesis was performed. Most included studies were observational or validation-focused, with very few RCTs, limiting the strength of evidence for clinical decision-making. The field is highly heterogeneous in design, population, and outcome measures, making direct comparisons between studies difficult. Only 5.7% of studies reported usability or clinical outcomes, meaning the practical impact on patient care or therapist workflow was rarely measured. Research is geographically concentrated in China and the USA, which may limit the generalizability of findings to other healthcare systems.
Declared interests
The authors declared no conflicts of interest with any financial organization. Funding for five authors was provided by the Italian Ministry of Health (Ricerca Corrente). The review was supported by CochraneRehab and the Italian Society of Artificial Intelligence in Medicine.
The easy way to misread this
Do not interpret the large number of AI studies as evidence that these tools are ready for clinical adoption. Most studies were small, lacked comparison groups, and did not test external validity or real-world clinical outcomes, so the current evidence base does not support changing practice based on AI performance metrics alone.
Summarised by AI from the full paper, without a clinician reviewing it. Check it against the source before it changes what you do. Read it on PubMed →