Researchers from the National Institutes of Health are using large language models to develop an artificial intelligence framework to streamline the clinical trial matching process and more quickly link potential volunteers to relevant trials listed on ClinicalTrials.gov.

Benchmarking its accuracy against three human clinicians, researchers have found that the tool, TrialGPT, achieved nearly the same level of accuracy, according to an NIH announcement this month.

WHY IT MATTERS

Because finding the right clinical trial for a patient is both time and resource-intensive, researchers at the National Library of Medicine and National Cancer Institute developed the TrialGPT framework to streamline it.

The new clinical trial-matching algorithm analyzes patient summaries for relevant medical and demographic information then identifies clinical trials for which a patient is eligible and excludes trials for which they are not. 

TrialGPT produces an annotated list of clinical trials – ranked by relevance and eligibility – that clinicians can use to discuss clinical trial opportunities with their patients. The AI tool also explains how a person meets the study enrollment criteria, which is critical to its efficacy. 

To assess how well TrialGPT predicted if a patient met a specific requirement for a clinical trial, the researchers compared the tool's results to those of three human clinicians who assessed more than 1,000 patient-criterion pairs, NIH said.

"Machine learning and AI technology have held promise in matching patients with clinical trials, but their practical application across diverse populations still needed exploration," Stephen Sherry, NLM's acting director, said in a statement.

The researchers also conducted a pilot user study and found that when clinicians used TrialGPT, they spent 40% less time screening patients but maintained the same level of accuracy.

Of note, TrialGPT relies on OpenAI’s GPT series LLMs such as GPT-3.5 and GPT-4 and the latter is closed-source and can only be accessed via commercial applications or API, the researchers said in their report.

For their study, published in Nature Communications and coauthored by collaborators from Albert Einstein College of Medicine, University of Pittsburgh, University of Illinois Urbana-Champaign and University of Maryland, College Park, the research team received an innovation award and will further assess the model’s performance and fairness in real-world clinical settings, NIH said.

THE LARGER TREND

Using AI to improve patient recruitment, retention and outcomes of clinical trials began before OpenAI launched its ChatGPT generative AI model. During the COVID-19 pandemic, oncology organizations sought ways to find patients across the country who would qualify for trials, even if they weren't physically there, through healthcare data.

In driving decentralized clinical trials, increased AI adoption helped to advance health equity and trial diversity, according to Jeff Elton, CEO of ConcertAI, a vendor of data and AI SaaS platforms for clinical trial optimization. 

"With integrated digital trials, clinical studies are integral to the care process itself, versus being imposed on it," Elton told Healthcare IT News.

"Trials don't need to place a higher burden on providers and patients than the standard of care." 

Reducing friction throughout the clinical trial lifecycle is critical to helping patients access trial therapies, according to Seth Howard, vice president of research and development at Epic. 

The electronic health record vendor implemented data-driven clinical trial matchmaking two years ago. Using its de-identified Cosmos data set, Epic allows providers that register for the service to match clinical trial opportunities from sponsors and a count of their organization’s eligible patients. 

Many health systems have also tested using analytical applications that can surface clinical trial opportunities for patients using their organizations' EHR data. In October, Microsoft announced new AI tools that will enable health systems to build their custom AI tools for many administrative needs – including clinical trial matching. 

Bias in AI, however, is still a concern for clinical outcomes. 

It can surface in any algorithm development pipeline and worsen healthcare disparities, Yale School of Medicine researchers said in new research published earlier this month.

ON THE RECORD

"Our study shows that TrialGPT could help clinicians connect their patients to clinical trial opportunities more efficiently and save precious time that can be better spent on harder tasks that require human expertise," Zhiyong Lu, NLM senior investigator and the study's corresponding author, said in a statement.

"This study shows we can responsibly leverage AI technology so physicians can connect their patients to a relevant clinical trial that may be of interest to them with even more speed and efficiency," Sherry added.

Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS Media publication.