ChatEHR: Integrating LLMs into Clinical Workflows at an Academic Medical Center
ChatEHR has been created by researchers to merge large language models (LLMs) with comprehensive patient timelines that extend over several years, thereby minimizing the challenges associated with manual data entry. This innovative system supports both automated processes—where prompts and data are fixed for specific tasks—and interactive functionality within the electronic health record (EHR) through a user-friendly interface. Over a span of 1.5 years, seven automation processes were developed, and 1,075 users received training to become regular users, performing tasks like pre-visit chart reviews, assessing transfer eligibility, tracking surgical site infections, and chart abstraction. ChatEHR transforms the application of LLMs into a key institutional resource, facilitating ongoing monitoring and assessment.
Key facts
- ChatEHR enables LLM use with the entire patient timeline spanning several years.
- The system supports automations and interactive use in the EHR via a UI.
- In 1.5 years, 7 automations were built.
- 1,075 users trained to become routine users of the UI.
- Use cases include pre-visit chart review, transfer eligibility screening, surgical site infection monitoring, and chart abstraction.
- ChatEHR reduces workflow friction from manual data entry.
- The system allows continuous monitoring and evaluation of LLM use.
- Access is granted after user-training.
Entities
Institutions
- arXiv