Mass General Brigham
Ambient AI Scribes Across Five Health Systems: A Multi-Site Study Finds Modest Time Savings
- EHR time: about 13 minutes less per day (3%)
- Documentation time: about 16 minutes less per day (10%)
- Visits: about 0.5 more per week; about $167 more revenue per adopter per month
- Largest reductions for clinicians using the scribe in more than half of visits
The Outcome
Adopters used the electronic health record about 13 minutes less a day and spent about 16 minutes less on documentation (relative drops of 3% and 10%), with about 0.5 more patient visits a week. The authors said time savings are unlikely to fully explain changes in burnout.
Key Metrics
- EHR time: about 13 minutes less per day (3%)
- Documentation time: about 16 minutes less per day (10%)
- Visits: about 0.5 more per week; about $167 more revenue per adopter per month
- Largest reductions for clinicians using the scribe in more than half of visits
The Problem
Documentation burden is a leading driver of clinician burnout, and health systems needed independent evidence on whether ambient AI scribes actually reduce it.
The Solution
Researchers from Mass General Brigham and UCSF followed ambulatory clinicians at five US health systems (Mass General Brigham, Emory, UCSF, Yale New Haven Health and UC Davis) for over two years, comparing 1,809 AI scribe adopters with other clinicians among 8,581 in total.
Key Takeaways: Mass General Brigham AI Results
- Company: Mass General Brigham (Healthcare)
- EHR time: about 13 minutes less per day (3%)
- Documentation time: about 16 minutes less per day (10%)
- Visits: about 0.5 more per week; about $167 more revenue per adopter per month
- Largest reductions for clinicians using the scribe in more than half of visits
- ROI multiple: Not publicly disclosed
Sign in free to read the full Mass General Brigham case study
A free AI Hub account unlocks these sections, plus downloadable templates and Academy courses.
- ROI & financial analysis
- Implementation journey & timeline
- Lessons learned & challenges
- Technical architecture & tech stack
- Governance & oversight
- Sources & code resources
Parts of this page were prepared with AI tools and reviewed and edited by a person, who is responsible for it. How we use AI