Using AI to Better Recruit Clinical Trial Participants

By DocWire News Editors - Last Updated: April 11, 2023

Researchers often struggle to find the right participants for their clinical trials. Many studies have very specific requirements, and it is imperative that these studies not only enroll enough participants but that these volunteers meet the proper inclusion criteria as well. To facilitate the recruitment process in clinical trials, researchers from the Cincinnati Children’s Hospital Medical Center have recently created an artificial intelligence (AI) platform that identifies eligible patients for trials based on Electronic Health Records (EHRs). Their work was published in JMIR Medical Informatics.

Advertisement

Lead investigator Yizhao Ni, Ph.D., claims that their AI system, the Automated Clinical Trial Eligibility Screener (ACTES), leverages EHR data to rapidly screen potential candidates. Though traditional methods of recruiting participants in busy medical environments such as the emergency department have been effective, they don’t always recruit enough individuals.

“Because of the large volume of data documented in EHRs, the recruiting processes used now to find relevant information are very labor-intensive within the short time frame needed,” said Ni. “By leveraging natural language processing and machine learning technologies, ACTES was able to quickly analyze different types of data and automatically determine patients’ suitability for clinical trials.”

Technology Behind AI Technology

ACTES functions using natural language processing, enabling it to comprehend human language as it analyzes a vast amount of data. The system rapidly extracts information such as patient demographics and other statistics from EHRs to compare to the set inclusion criteria for a given clinical trial. This advanced AI also pulls unstructured information from the physician’s clinical notes as well, including symptoms, treatments, and diagnoses.

The AI is also equipped with machine learning capabilities that allow it to automatically learn and improve its processes over time without further programming. This component also enables ACTES to learn from previous trial enrollments to enhance its selection skills using AI algorithms.

Background of the ACTES Study

ACTES was previously tested in a retrospective study published in 2015, but this new research evaluated the system in real-time within a busy emergency department. This was done at the pediatric emergency department at Cincinnati Children’s Hospital Medical Center. ACTES continuously analyzed EHR data for these patients and recommended them for any of six potential clinical trials. Patient eligibility data was presented to the personnel in charge of the clinical trials in real-time on a dashboard.

This study found that compared to the manual screening process, ACTES cut patient screening time down by 34% in the recruitment process. In addition, patient enrollment increased by 11.1%, the number of patients screened by 14.7%, and the number of patients approached by 11.1%.

This use of this AI in the clinical environment required collaboration between the research scientists, software developers, information technicians, and the clinical staff.

“Thanks to the institution’s collaborative environment, we successfully incorporated different groups of experts in designing the integration process of this AI solution,” Ni concluded.

The researchers noted that a limitation to the study is that it only recruited patients from one center for a small number of different trials. It could become more challenging for ACTES to identify potential candidates if the number of clinical trials it had to match increased. Ni and colleagues also noted several issues regarding the system’s accuracy in interpreting data, but they claim these issues will be resolved in future studies.

Advertisement