The Next Cure, Faster: Why we Built Brim Analytics
February 13, 2025
Live-Saving Medical Research is Powered by Data
Every family has been touched by disease—whether it’s cancer, heart disease, or a rare genetic disorder. We’re all waiting for the next medical breakthrough. But what if the biggest obstacle to that breakthrough isn’t science itself, but data trapped in electronic notes?
Right now, only 3-5% of eligible patients ever get matched to a clinical trial. Not because there aren’t enough trials—but because the data researchers need is buried in unstructured medical records, making patient recruitment slow, expensive, and inefficient.
The same problem slows down medical registries, quality improvement programs, and other life-saving research.
And the culprit? Manual chart abstraction.
Chart Abstraction Is a Major Bottleneck to Good Medical Data
One of the biggest barriers to improving patient outcomes and accelerating research is chart abstraction—the process of extracting data from unstructured clinical notes. It’s a time-consuming, expensive, and highly manual task.
- Hospitals, medical schools and research teams spend millions of dollars a year hiring trained professionals to do this work.
- Clinical trials fail to enroll thousands of eligible patients—simply because their data are trapped in electronic notes, and there’s not enough abstractor bandwidth to get it to the researchers who need it most.
- The size and detail of medical registries is limited by the painstaking process of manually curating patient records, limiting the downstream learnings.
The irony? We have the computational power to analyze and interpret medical data at a much bigger scale than we are now. What we don’t have is easy access to that data in a flexibly structured way.
What if we could change that? What if we could find the next cure faster by drastically reducing the time it takes to:
- Identify trial candidates in real-time.
- Detect co-morbidities and genetic risk factors at scale.
- Build better medical registries with high-quality, structured data.
A Vision for the Future of Chart Abstraction
Dr. Dan Fabbri has spent years working on solutions to the chart abstraction problem in different applications and honing his vision for the future of chart abstraction.
That vision: to radically improve the speed and scale of chart abstraction with a solution that is also:
- Privacy-First. Analytics and AI products are often data-hungry because data helps them improve the software over time. But medical researchers using patient charts need solutions that put privacy and security first.
- Accurate. We need speed to improve the industry overall, but accuracy is critical. We simply can’t afford to get clinical insights wrong.
- Democratized. Solutions in the space need to work not just for engineers, but for researchers, doctors, and clinical staff.
Dr. Fabbri has built different approaches over that time, from crowdsourcing to NLP-powered search, with each approach coming a bit closer to his vision. In the last 2 years, large language models (LLMs) have become powerful and popularized. Suddenly, the technical apparatus to scale chart abstraction securely, accurately, and at unprecedented speed exists.
Brim Is That Future
The Brim chart abstraction tool is the solution we always envisioned.
- LLM-powered chart abstraction speeds up medical research by structuring unstructured notes.
- Human-in-the-loop validation ensures accuracy and trustworthiness, with raw text from the notes supporting every data point. Today’s manual chart abstractors can be 10 times as productive with Brim as a tool.
- Privacy-first infrastructure means researchers retain control of medical records. Our Bring your Own LLM design allows institutions to work with LLMs that won’t share their data.
- Designed for non-technical users so medical researchers can work with the data without a deep knowledge of prompt engineering, LLMs, or data pipelines.
The future of medical research doesn’t have to be bottlenecked by slow, expensive data abstraction. Brim is here to change that.
Stop letting manual data abstraction slow down your research. See how Brim can cut your workflow from hours to minutes—without compromising accuracy.