Evidence on artificial intelligence-assisted clinical documentation and healthcare workers' emotional wellbeing at...

This scoping review mapped evidence on artificial intelligence-assisted clinical documentation for healthcare workers' emotional wellbeing at work, including tool types, reported favorable, adverse, or mixed findings, and evidence gaps. We also considered how these tools may shape documentation-related work demands, autonomy, clinical voice, patient connection, and broader occupational wellbeing.

We searched PubMed, Web of Science, Embase, CINAHL, and PsycINFO from database inception to March 16, 2026, to identify studies. The review was conducted in accordance with the Joanna Briggs Institute methodological framework and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews guidelines. Two reviewers independently conducted study selection and data extraction. The findings were synthesized using descriptive and narrative approaches, and the included studies were critically appraised using the Mixed Methods Appraisal Tool 2018.

Thirty-five studies met the inclusion criteria. The evidence base was highly concentrated: 31 of the 35 studies were conducted in the United States, and 30 evaluated ambient artificial intelligence scribe tools. Overall, 23 studies reported predominantly favorable findings, while 12 were classified as reporting mixed findings. The most frequently reported benefits included reduced documentation burden, decreased cognitive load, improved work satisfaction, and better perceived patient connection. Outcomes related to burnout were highly variable. Mixed findings were primarily associated with implementation barriers, the need for editing, accuracy concerns, and challenges related to preserving clinical voice and professional autonomy.

Current evidence indicates that artificial intelligence-assisted clinical documentation is mainly associated with clinician-reported relief in documentation-proximal strain, especially perceived documentation burden and cognitive load. However, the evidence remains concentrated in early evaluations of ambient artificial intelligence scribes in United States healthcare settings and should not be generalized to all documentation artificial intelligence tools or health systems. Findings for broader emotional wellbeing outcomes, including burnout, remain limited and mixed. Given the methodological concerns identified in the Mixed Methods Appraisal Tool appraisal, these findings should be interpreted as reported associations and perceived changes rather than causal evidence. Future studies should use longitudinal, multicenter designs and validated wellbeing measures to assess durability, safety, and longer-term occupational outcomes.

https://osf.io/m8e9v.