Evaluating Ambient AI for Clinical Documentation: A Mixed-Methods Study
Translated title
Evaluering af Ambient AI til klinisk dokumentation: Et mixed-methods studie
Authors
Term
3. term
Education
Publication year
2025
Submitted on
2025-12-16
Pages
22
Abstract
Background & Objective: Documentation in electronic health records increases clinician workload and burnout. Ambient AI scribes may re-duce this burden by generating draft notes, but challenges remain. This study investigates how ambient AI scribes affect documentation work in clinical practice. Methods: A mixed-methods explanatory sequential design was used. Two clinicians, a psychologist and a nurse, participated. Quantitative data included documentation time and error rates, analysed using Mann-Whitney U tests and Pearson’s and Spearman’s correlations. Qualitative data included a semistructured interview, analysed thematically follow-ing Braun and Clarke. Results: Quantitative analyses showed increased documentation time for the nurse (7.5 vs. 21.5 min, p = 0.024), while no difference was found for the psychologist (62.0 vs. 86.4 min, p = 0.190). A learning effect was observed only for the psychologist (ρ = –0.689, p = 0.040, 95 % CI [–0.932, –0.022]). Mean error rates were higher for the psychologist (34.11 (SD 12.33)) than the nurse (9.33 (SD 3.50). Error rate was not associated with documentation time or number of sessions. Qualitative analyses indicated that ambient AI was intuitive and helpful for structuring draft notes, however frequent errors, missing infor-mation, and hallucinations limited trust and prevented time savings. Conclusion: Ambient AI did not improve documentation efficiency, was associated with ambivalent user satisfaction, and did not demon-strate sufficient effectiveness. Further research is necessary due to study limitations.
Documents