Manual chart abstraction is time-intensive, difficult to scale, and often a bottleneck for research. BRIM is designed to change that. 

UCSF Academic Research Services (ARS) is excited to introduce BRIM, a new AI-guided chart abstraction platform now available for pilot testing through June 30. 

BRIM is an institutionally hosted, web-based platform created at Vanderbilt University that enables teams to transform unstructured clinical notes into structured, analyzable data—combining AI-assisted abstraction with human review to support high-quality research workflows. 

What Makes BRIM Different 

BRIM is built for researchers and other clinical data users — not machine learning engineers. It allows you to: 

• Define complex variables without writing code  

• Apply AI-assisted abstraction across clinical notes and structured data  

• Review, validate, and refine outputs with full transparency  

• Export structured datasets for downstream analysis  

Unlike traditional NLP pipelines or manual abstraction workflows, BRIM keeps humans in the loop, ensuring both scalability and data quality. 

Designed for Real Research Workflows 

BRIM supports a wide range of use cases across UCSF, including: 

• Clinical research chart abstraction  

• Registry development and maintenance  

• Cohort identification and feasibility analysis  

• Quality improvement and operational analytics  

• Retrospective studies using clinical notes  

No specialized machine learning expertise is required to get started. 

🚀 UCSF BRIM Pilot Now Open (Through June 30) 

BRIM is available for a limited pilot period through June 30, and your participation is critical. 

We encourage you to: 

• Use BRIM in an active or upcoming project  

• Assess its impact on speed, scalability, and data quality  

• Share feedback on what works—and what doesn’t  

Pilot projects include free usage of up to 5M tokens of processed text. This is typically sufficient to test BRIM with a small set of notes (e.g., 10–30 notes) and a focused set of variables. 

For larger or production use, projects will incur Versa API usage costs, and users will be required to provide billing details. During the pilot period, BRIM access is free, and only API usage is billed. 

Your feedback during this pilot will directly inform decisions about BRIM’s long-term adoption at UCSF. Research IT leadership will use input from this pilot to evaluate long-term availability and integration into UCSF’s data infrastructure. 

Get Started 

Getting started with BRIM is straightforward: 

1. Request access via the UCSF Brim Intake Form  

2. Review the startup guide and tutorials  

3. Begin defining variables and uploading data for abstraction  

BRIM supports secure use of UCSF institutional data, including P3/P4 data via approved pipelines. Use of such data for research requires appropriate IRB approval and adherence to UCSF data governance policies and guidance for responsible use of AI tools.  

Review applicable governance and AI guidance before using BRIM or other tools with non-UCSF P3/P4 data. 

Interested in Long-Term Use? 

At the end of the pilot period, surveys will be sent to all users, which will inform decisions about BRIM’s long-term availability. 

Departments and research groups interested in continued use beyond the pilot are encouraged to contact ARS Information Commons team at [email protected] to discuss sustainability and support models. 

Learn More & Get Support 

• Review documentation and tutorials  

• Contact ARS Information Commons team at [email protected] for consultation and onboarding support