National Institutes of Health Develops New Algorithm to Accelerate Clinical Trial Enrollment and Research
Volunteer recruitment is one of the biggest challenges biopharmas face. Even the best trial designs aren’t immune. Consider that up to 85% of clinical trials struggle to retain or recruit enough patients. Although various factors contribute to this discrepancy –– including study protocols, fear of side effects, and lack of awareness –– one of the biggest obstacles is eligibility criteria. The good news is that may be changing soon, thanks in part to artificial intelligence.
National Institutes of Health (NIH) Introduces TrialGPT
Researchers at NIH have developed a new artificial intelligence algorithm designed to speed up the trial recruitment process. Dubbed, TrialGPT, the algorithm matches potential volunteers with relevant clinical trials listed on ClinicalTrials.gov. Even better, though, is the algorithm works!
A study published in Nature Communications found that TrialGPT could:
- Successfully identify relevant trials for eligible volunteers
- Provide a summary that clearly explains how the volunteer meets the criteria
- Help clinicians navigate the array of clinical trials available to their patients
The study authors concluded that proper implementation could lead to improved clinical trial enrollment and faster progress in medical research.
Developing TrialGPT: A Peek Behind the Scenes
Using artificial intelligence for recruitment purposes may sound futuristic, but the science is sound. Two teams of researchers –– one from the NIH’s National Library of Medicine and the second from the National Cancer Institute –– developed an innovative framework that allowed TrialGPT to speed up the clinical trial matching process. Here’s a closer look at how it works:
Step One: Patient Summary
First, TrialGPT reviews a patient summary containing relevant medical and demographic information.
Step Two: Database Comparison
Once that’s finished, the algorithm reviews relevant clinical trials on ClinicalTrials.gov. It highlights trials where patients might be eligible while excluding others in which the patient wouldn’t meet the necessary requirements.
Step Three: Annotated List of Eligible Trials
After sorting through all that information, TrialGPT provides a complete list of clinical trials for the patient, ranked by relevance and eligibility. The list also includes information on each trial, explaining in detail how the patient meets the specific enrollment criteria.
The National Library of Medicine’s acting director, Stephen Sherry, PhD says “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.”
TrialGPT vs. Human Clinicians: A Comparison
These initial findings are exciting, but researchers wanted to confirm that TrialGPT could do the work of human clinicians. To gain more information, the investigators compared how well the AI algorithm could predict if a patient met specific clinical trial requirements compared to three human clinicians.
The experiment evaluated more than 1,000 patient-criterion pairs. At the end of this process, the researchers concluded that TrialGPT achieved almost the same level of accuracy as the clinicians. Further, a complimentary study found that clinicians who used TrialGPT in the recruitment process spent 40% less time screening patients while maintaining the same level of accuracy.
What To Expect Going Forward
TrialGPT isn’t available for broad adoption yet, but the research team responsible for the above findings was recently selected for The Director’s Challenge Innovation Award. The funds awarded through this process will help the team further assess the AI model’s performance and fairness in real-world clinical settings.
It’s an exciting time to be involved in the industry! We’ll be keeping close tabs on the continuing research and provide more updates as we get them. For more content like this, check out our blog and follow us on LinkedIn. We’d love for you to join the conversation.
