Rashid awarded grant to develop AI tool that recommends optimal clinical trials to pancreatic cancer patients

September 24, 2024

By Tyler Rice, UNC Lineberger Pancreatic Cancer Center of Excellence

Dr. Naim Rashid

Naim Rashid, PhD, has received a two-year, $311,000 Department of Defense Pancreatic Cancer Research Program-focused pilot award to build an artificial intelligence (AI) tool that generates personalized clinical trial recommendations for patients with pancreatic ductal adenocarcinoma (PDAC).

Rashid, an associate professor of biostatistics at the UNC Gillings School of Global Public Health and a member of UNC Lineberger Comprehensive Cancer Center and the Pancreatic Cancer Center of Excellence (PCCE), will create the tool by gathering data from PDAC patients at UNC Health and linking it to PDAC trial listings on ClinicalTrials.gov. The project’s goal is to enhance patient knowledge and awareness of clinical trials, promote patient self-advocacy through discussions about trials with oncology providers, reduce racial and geographic disparities in trial participation and, ultimately, increase enrollment in trials.

Significant advances in care have helped double the five-year overall survival rate of pancreatic cancer in the past two decades, largely in the form of new treatments and practices developed in clinical trials.

The Institute of Medicine recommends that every patient has access to cancer clinical trials, but not all PDAC patients have the same level of access. For example, there are well-documented racial disparities in clinical trial participation, attributed partially to inequitable recruitment practices and varying awareness of eligible trials among underrepresented populations. Furthermore, PDAC clinical trials are being introduced increasingly quickly on databases such as ClinicalTrials.gov, and navigating eligibility criteria beyond disease stage and type has become more complex, placing greater burden on patients and physicians to find suitable trials.

Through this innovative project, Rashid and his team aim to empower PDAC patients with tailored insights about trials relevant to their specific diagnosis. Working with a physician-led team to determine patient eligibility for trials, Rashid’s group will “fine-tune” an existing, pre-trained large-language model (LLM) by matching prior data from PDAC patients at UNC to downloaded trials.

They will then develop a personalized recommendation system that uses LLM and operates like an AI chatbot to provide high-quality ranked matches. The chat function will let patients ask questions and give preferences in natural language to receive answers about the trials that best suit their needs. The LLM will be further fine-tuned with new trials and retrospective patient data, as well as feedback from providers, to keep recommendations up to date.

Rashid’s group will also create an easily accessible mHealth application that connects to the AI tool to suggest eligible trials, helping patients and health care providers to more easily navigate a rapidly evolving database. The mobile app will share reminders and educational materials that might be useful to both parties, and it will be updated with newly posted trial data. Evaluating the app in real-time and measuring its impact, Rashid’s group will follow the enrollment of an additional 100 patients at UNC Health’s hospitals and clinics and make iterative adjustments to improve recommendations.

Rashid believes their intuitive, app-based AI tool will simplify the trial discovery process by removing guesswork from the search, clarifying the trial landscape and facilitating discussion between patients and their care team.

“Clinical trials are critical to advancing new treatments and therapies to cancer patients, and we hope that our AI mHealth app will help level the playing field for PDAC patients to self-advocate and find trials they are eligible for,” Rashid said. “In the future, we hope to expand the scope of this app to extend to other cancer types using similar methods.”

The mHealth initiative stems from work led by UNC Lineberger’s Marjory Charlot, MD, MPH, MSc, including her Sisters CREATE study, and is also part of the PROCLAIM Study, which is recruiting patients. PROCLAIM examines whether mHealth apps can help break down barriers to diverse recruitment in PDAC trials, seeking to better inform minority patient populations of relevant clinical trials and encourage patients to ask their providers about trials that might improve their outcomes. The study is led at UNC Health by Charlot, Jen Jen Yeh, MD, director of the PCCE and professor of surgical oncology and pharmacology, and Ashwin Somasundaram, MD, assistant professor of medical oncology; and at the Medical College of Wisconsin by Ugwuji Maduekwe, MD, MMSc, MPH, FACS, FSSO. Charlot, Yeh and Somasundaram are also members of UNC Lineberger.

Other key personnel on Rashid’s grant include Yeh; Michael Kosorok, PhD, W. R. Kenan, Jr. Distinguished Professor of biostatistics and professor of statistics and operations research and UNC Lineberger member; and Summer Choudhury, MPH, of the NC TraCS Institute.

Contact the UNC Gillings School of Global Public Health communications team at sphcomm@unc.edu.