1228 Clinic to Couch: A Quality Improvement Project Analyzing the Effect of Artificial Intelligence Scribes on Documentation Burden in Our Sleep Center

Abstract Introduction Artificial intelligence (AI) tools are increasingly used in clinical practice to reduce documentation burden and improve workflow efficiency. This quality improvement (QI) project evaluated the impact of Ambience, an ambient AI scribe, on documentation-related metrics and adoption patterns among sleep clinicians over a short observational period. Methods Only providers who completed training and actively used Ambience were included. Data was analyzed for 27 providers in the Cleveland Clinic Sleep Disorders Center who utilized Ambience. Provider-level adoption data (training status, Ambience encounters, first/last use dates) were obtained from institutional records. Documentation metrics included documentation as a percent of clinic time and documentation time outside scheduled hours, were extracted from Ambience for 4 users who used Ambience for more than 180 days. Data was collected for four months pre-implementation (December 2024–March 2025) and two months post-implementation (May–June 2025), excluding April as a transition month. Results Review of data showed individual Ambience encounter volumes ranged from 236 to 1,325, with active use spanning 15–231 days. Among 27 providers reviewed, 30-day utilization of Ambience was 77.78 % (21/27) and 90-day utilization was 70.37% (19/27). Review of documentation metrics for 4 providers with more 180-day usage showed a decrease in the Documentation as percent of clinic time, with a decrease in mean aggregate from 31.5 % (pre-Ambience) to 23.09 % (post-Ambience). While the aggregate mean of documentation time outside scheduled hours increased from a mean aggregate of 4.07 % (pre-Ambience) to 4.9 % (post-Ambience). Conclusion The successful adoption and active usage of Ambience indicate the tool's ease of implementation. Usage variability at 30 and 90 days highlights the need for continued interaction and workflow adjustments to improve transition. While reduced in-clinic documentation suggests improved workflow efficiency and higher patient interaction, increased after-hours work warrants investigation into contributing variables, such as the time required for reviewing and editing AI-generated notes. Larger follow-up studies are required to confirm these findings. Overall, AI scribes show promise for boosting documentation efficiency and patient-centered care. Support (if any)