Researchers at Severance Hospital, Yonsei University College of Medicine have developed an AI model that can predict the risk of early liver metastasis in pancreatic cancer patients using routine blood test data. The model, called LiMPC, is now being translated into an online clinical tool to help physicians assess metastasis risk without relying solely on imaging.
LiMPC was trained on data from 2,657 pancreatic cancer patients and externally validated on an additional 272 patients across five hospitals in South Korea. Results published in BMC Cancer showed the model correctly identified about 81% of patients with liver metastasis and demonstrated strong rule-out capability, with 87% of low-risk patients confirmed to be metastasis-free.
The tool addresses a key clinical challenge, as early metastasis in pancreatic cancer is often difficult to detect through imaging alone. Researchers say the model could serve as a complementary method to standard diagnostics, particularly in early-stage cases where metastases are not easily visible. Because it relies only on widely available blood tests, the approach may also be valuable in resource-limited settings without access to advanced imaging technologies.
The team plans to launch an online calculator based on the model, enabling clinicians to quickly estimate metastasis risk during routine care. Further studies are underway to refine the model and support integration into clinical workflows. The development aligns with broader efforts across Asia to use AI for earlier detection of pancreatic cancer and improve diagnostic accuracy.
Click here to read the original news story.