At the Mayo Clinic, the Nvidia SuperPOD is powering AI to a breakthrough in pancreatic cancer diagnosis.
Pancreatic cancer is the 11th most common cancer in the United States, but it is the third deadliest. Half of all pancreatic cancers are first diagnosed when a patient is already at stage four and the five-year survival rate is just 13%. But with AI at the Mayo Clinic, it’s a new ballgame.
“The latest generation DPE that we just installed in our SuperPOD, working with Nvidia, allows us to go much more efficiently building models. We can work with much larger amounts of data to do those models,” said Dr. Matthew Callstrom, director of AI strategy at the Mayo Clinic.
The AI developed models that can detect cancer that the human eye and other machines cannot. The technology still has to go to clinical trials, but so far, tests have shown AI can detect cancers 438 days earlier than conventional methods, and that means catching the cancer earlier.
“To compare how well it works, I am about 50% being able to pick up these early-stage cancers, whereas this model that developed can actually pick it up 97% of the time,” said Callstrom.
In the SuperPOD, doctors have trained the AI by having it analyze more than a million slides from cancer patients to develop its ability to see what humans can’t.
“Between the AI model running on the CT scan and the radiologist, we really have a much better opportunity to detect those early-stage cancers,” said Callstrom.
