Quantcast
Channel: News
Viewing all articles
Browse latest Browse all 184

The Evolution of LLMs in Healthcare

$
0
0
The Evolution of LLMs in Healthcarejordan.scott_xWTB

Just six months after OpenAI released the large language model ChatGPT, three Stanford University physicians penned a commentary wondering if LLMs  “will reshape modern medicine.” Writing in 5Internal Medicine in April 2023, they continued, “Good or bad, ready or not, Pandora’s box has already been opened.”

LLMs, defined by Gartner as artificial intelligence models“trained on vast amounts of text to understand existing content and generate original content,” have found many use cases in healthcare in less than two years. Some center on facilitating communication with patients, while others attempt to analyze large sets of unstructured data for clues about a patient’s condition or to determine appropriate billing codes.

LLMs in healthcare are undergoing near-constant change as developers refine their models to improve accuracy and remove bias, and health systems determine the workflows that are appropriate for using AI. As a second JAMA paper from Stanford concluded, the industry needs to ensure models make medical professionals more productive, rather than simply automating tasks they already know how to do.

PREPARE: Expert guidance helps healthcare organizations achieve meaningful transformation with AI.

Five Key Roles for LLMs in Healthcare

There are two general categories of LLMs, according to the second Stanford piece. One category is trained on medical documents, ranging from progress notes to medical literature, and is typically deployed to summarize lengthy records or answer clinical questions. The second is trained on structured medical codes, generates a “high-dimensional vector representing the patient’s medical record,” and aims to predict medical events such as readmissions or lengthy hospital stays.

“LLMs excel at summarizing information accurately and even suggesting decisions based on their analysis,” says Venky Ananth, executive vice president and global head of healthcare at Infosys.

As a result, LLMs are well positioned to address two valuable clinical use cases that otherwise involve heavy reading: Disease management and prior authorization. “We’re seeing high adoption in the prior authorization space, as this often involves wading through massive amounts of clinical data, physicians’ notes and lab results spread across dozens of pages.”

A 2023 paper in Communications Medicine listed three additional use cases for LLMs in improving patient care:

  • Better communication between patients and provider organizations, especially after business hours and for difficult conditions that carry a social stigma
  • Fast and accurate translations of medical information into “plain, everyday language” as well as languages other than English
  • Reporting, documentation and other administrative requirements, which researchers estimate can occupy at least 25 percent of a clinician’s workday

Click the banner below for expert guidance to help optimize your IT operations.


Viewing all articles
Browse latest Browse all 184

Trending Articles