Researchers at UC San Diego Health have found that by using AI they could develop more timely and efficient assessments of care provided to patients with severe sepsis in the emergency department.

The study results, published online in JAMA Network Open, used AI and large language models (LLMs) to automatically assess complex measurements of care and delivered targeted feedback for providers in the hospital.

“Medicine can learn a lot from professional athletics, where every player knows their performance statistics almost immediately, and that feedback changes how they train and perform,” said Gabriel Wardi, M.D., co-corresponding author of the paper, in a statement. 

“Rarely do physicians get that same type of quick, individualized feedback, even for conditions as time-sensitive as sepsis. By measuring performance in near-real time, we can turn quality reporting from a retrospective administrative exercise into something that actually helps physicians improve care,” added Wardi, an emergency and critical care medicine physician at UC San Diego Health and chief of the Division of Critical Care in the Department of Emergency Medicine at UC San Diego School of Medicine.

In partnership with the Joan & Irwin Jacobs Center for Health Innovation at UC San Diego Health, study researchers found that LLMs can perform accurate abstractions for complex quality measures, particularly in the challenging context of the Centers for Medicare & Medicaid Services (CMS) SEP-1 measure for severe sepsis and septic shock.

Traditionally, the clinical review process for SEP-1 involves a 63-step evaluation of extensive medical charts, requiring months of effort from multiple reviewers for a few patient cases. This study found that LLMs can dramatically reduce the time and resources needed for this process by accurately scanning hundreds of patient charts and generating critical contextual insights in seconds, oftentimes while the patient is still being cared for in the hospital. Once the charts are automatically reviewed through the LLM, a notification is sent to clinical leadership in the Emergency Department for further evaluation and then disseminated to the medical teams providing treatment to the patient with sepsis. Through this effort, physicians are given near real-time feedback on their patient case and, when necessary, provided recommendations for meeting SEP-1 guidelines.

“By using AI to quickly assess sepsis quality care measures, we are able to provide guidance to our care teams in the most teachable moment,” said Karandeep Singh, M.D., study co-author, and chief health artificial intelligence officer at UC San Diego Health, in a statement. “In turn, this has resulted in improved compliance with national sepsis quality measures and helps our teams consistently improve upon the care they provide to the communities we are proud to serve each day,” added Singh, who also is Joan & Irwin Jacobs Endowed Chair in Digital Health Innovation at UC San Diego School of Medicine.

Other key findings of the study found that LLMs can improve efficiency by correcting errors and lowering administrative costs by automating tasks, which are scalable across various healthcare settings.

“Using small, privacy-preserving language models allows for rapid and actionable insights distilled from large amounts of documentation in medical charts,” said Aaron Boussina, Ph.D., first author of the paper and affiliate faculty at the Joan & Irwin Jacobs Center for Health Innovation at UC San Diego School of Medicine, in a statement. “This seamlessly embeds best practices in the care delivery process.”