���Ҵ�ý

Aditya Chakraborty, Ph.D., strives to translate data into hope for people with pancreatic cancer. Pancreatic cancer has the highest mortality rate of all major cancers, with a five-year relative survival rate of just 13%, according to the American Cancer Society. Early detection is difficult, because symptoms usually appear only after the disease has advanced and when surgery is no longer an option. 

Dr. Chakraborty is an assistant professor in the Department of Epidemiology, Biostatistics and Environmental Health in the Joint School of Public Health at Old Dominion University, in partnership with Norfolk State University. He said a mentor, who had survived three different types of cancer, inspired him to pursue this research during graduate school. 

He and his team, which included researchers from the University of Florida, analyzed thousands of pancreatic cancer patient records to predict risk categories. Utilizing artificial intelligence (AI), they tested different machine learning models using anonymous patient data from the National Cancer Institute’s database. 

“When it comes to AI or advanced modeling, you need to be careful, because not all models give you the desired accuracy,” Dr. Chakraborty said. The team refined the models by weighing the factors that most influenced risk and treatment. They then validated the models using a second patient dataset to ensure consistent results. This work led to the creation of a Survival Indicator, or SI — a tool that predicts survival time from diagnosis and shows whether a patient’s outlook is likely to improve based on treatment decisions and individual risk factors. 

“We were able to better understand what treatments — chemotherapy, radiation or a combination of both — worked best,” Dr. Chakraborty said. 

The SI can also compare survival patterns across patient subgroups based on race, ethnicity, age, gender, cancer stage, treatment and other risk factors. It helps identify whether survival outcomes for a specific group are likely to improve, decline or remain stable over time. For example, it could be used to examine mortality outcomes related to hypertension, cancer or other chronic diseases for Black patients in Norfolk, Virginia, compared with other groups.

Although the tool was developed using mortality data from cancer patients in a national dataset, the framework can be applied more broadly to other chronic conditions within a specific local setting. It’s worth noting that the pancreatic cancer death rate in Norfolk is above the national average.

The next step is using the SI to address these disparities. Working with Julius O. Nyalwidhe, Ph.D., director of the George L. Wright, Jr. Center for Biomedical Proteomics and professor of biomedical and translational sciences at Macon & Joan Brock Virginia Health Sciences Eastern Virginia Medical School at Old Dominion University, Dr. Chakraborty and his team submitted a grant proposal to develop advanced, data-driven tools that identify and map cancer disparities across Hampton Roads. 

“What I am most excited about is the ability to transform advanced data science into practical, community-based solutions,” Dr. Nyalwidhe said. “I hope this work will give Hampton Roads the predictive tools, risk maps and partnerships needed to meaningfully reduce cancer disparities.” 

The project aims to integrate machine learning, biomarker analytics and community engagement to identify high-risk neighborhoods and support targeted cancer screening and prevention strategies. 

“Machine learning is unlocking insights from thousands of patient cases that no single physician could see alone,” Dr. Chakraborty said. “Our hope is that we will be able to detect pancreatic cancer earlier and treat it in a tailored and precise way — one patient at a time.”