AI could turn every surgical patient into an 'information donor'

"It would be like having all of the wisest people in the room do a tumor board.”

By
Meg Barbor
September 15, 2026
5 min

MDSpire News

Overview

Large language models (LLMs) can enhance data extraction from electronic medical records. This approach aims to improve the volume of patient data available for clinical research.

Background

Clinical research often relies on a limited subset of patient data, which can lead to biased findings. The use of LLMs to extract structured data from electronic medical records presents an opportunity to include a broader patient population.

Data Highlights

No numerical data presented in the article.

Key Findings

LLMs can extract structured data from narrative clinical notes, significantly increasing the volume of usable patient data.
Agreement between LLM-generated data and manually collected data exceeded 98% for several key metrics.
Manual data collection traditionally involves significant time and expertise, limiting the number of patient records reviewed.
Only 1% to 5% of patients typically contribute data for clinical research.
LLMs could reduce the error rate in outcomes collection, which is estimated at 5% to 7% with traditional methods.