
Cancer of the pancreas has one of the lowest survival rates, and delays in care only worsen a patient’s chance of survival. Unfortunately, treatment delays are more likely to affect certain groups of patients, including those that typically struggle to navigate the healthcare system. Programs that are led by nurses to help guide patients through cancer treatment, called nurse navigation programs, can help improve cancer care and outcomes for all patients – but often these programs aren’t available to patients until after they have received a confirmed diagnosis. For many pancreatic cancer patients, this comes too late. If a nurse navigation program for pancreatic cancer began earlier, it could significantly reduce delays in diagnosis and care, thereby improving patient outcomes. This is exactly what Division of Surgical Oncology Assistant Professor Nabeel Zafar, MBBS, MPH aims to test with a new two-year, $311,000 pilot grant from the Department of Defense Pancreatic Cancer Research Program.
“Artificial intelligence (AI) technology makes it much easier and faster to scan radiology reports for pancreatic abnormalities. We believe that using this technology could help us identify patients at a much earlier stage in their diagnostic journey, which would allow nurse navigators to start working with patients much sooner,” explained Zafar. “But to design an AI-assisted program, we first have to understand how best to incorporate this technology into the workflow of existing nurse navigation programs, and then pilot test it to ensure it is both feasible and effective.”
Zafar’s research team, including experts from the UW Department of Medical Physics, the Department of Biostatistics and Medical Informatics, and the Wisconsin Surgical Outcomes Research program, will start by evaluating and testing different large language models so they can identify the best AI model to scan radiology reports for suspicious pancreatic lesions. In addition, they will meet with nurse navigators, oncologists, radiologists, clinic managers, and other key partners at UW Health to capture and map their experiences and points of interactions with patients who are beginning their pancreatic cancer diagnosis and treatment journey. With the help of these key partners, they will identify where in the clinical workflow an AI-assisted nurse navigation program could be incorporated. They will then pilot test this program over a 6-month period at UW Health.
“Piloting this program at our institution could immediately improve the clinical care of the roughly 400 patients we see with pancreatic cancer each year,” said Zafar. “In addition, our local experience could be used by healthcare systems across the country to optimize pancreatic cancer diagnosis and care, potentially improving the lives of the more than 33,000 U.S. patients who currently experience delays in pancreatic cancer diagnosis every year.”