Early detection for some cancers can come up to three years before a traditional diagnosis with the aid of artificial intelligence, researchers have found according to a May 8 news release from Harvard Medical School.
An AI tool was successful at detecting pancreatic cancers up to three years before diagnosis by examining high-risk patients' medical records and comparing them with data on disease trajectories. The tool was trained on a dataset of 9 million patient records and built through a collaboration between Harvard Medical School, the University of Copenhagen, VA Boston Healthcare System, the Dana-Farber Cancer Institute and the Harvard T.H. Chan School of Public Health.
"Our results indicate that using the time sequence in disease histories as input to the model, rather than just disease occurrence at any time, improves the ability of AI methods to predict pancreatic cancer occurrence, especially for the highest-risk group," researchers wrote of their findings published May 8 in Nature.
Pancreatic cancer is an especially deadly disease with just a 12 percent five-year survival rate. This year, it is expected that 64,050 U.S. adults will be diagnosed with the disease and of those, 50,550 will die from it. Diagnoses numbers are projected to rise in coming years, according to Harvard Medical School.
A tool that can provide significantly earlier detection could shape better outcomes for patients across the board, the researchers note.
"AI-based screening is an opportunity to alter the trajectory of pancreatic cancer, an aggressive disease that is notoriously hard to diagnose early and treat promptly when the chances for success are highest," Søren Brunak, PhD, coauthor of the research and director of research at the Novo Nordisk Foundation Center for Protein Research at the University of Copenhagen said in the Harvard news release.