Deep Bio validates AI's role in prostate cancer prognosis

The results of a large-scale prostate cancer study, conducted jointly by Deep Bio—an AI-based digital pathology diagnosis solution company—and Pusan National University College of Medicine, were published on March 31 in Scientific Reports, an international journal by the Nature Publishing Group.

The study analyzed a total of 29,646 digital pathology whole-mount image slides from prostate cancer patients who had undergone radical prostatectomy. The findings demonstrated that Deep Bio’s AI-powered pathology image analysis algorithm is clinically valid for Gleason grade classification and tumor volume quantification in prostate cancer, and provided meaningful results for prognosis prediction, the company said.

Notably, an extended model combining the percent tumor volume (PTV) calculated by Deep Bio’s algorithm with an existing prognostic model (CAPRA-S Score) showed significantly improved predictive power for biochemical progression-free survival (BPFS) compared to the existing model alone (p=0.006). This suggests that AI pathology analysis can support clinical decision-making as a complement to conventional pathology diagnosis.

“This study is an important demonstration that our AI technology can analyze prostate cancer pathology data at scale and improve prognostic accuracy,” said Kim Sun-woo, CEO of Deep Bio. “By integrating AI into digital pathology workflows, we expect to enable more quantitative and consistent data-driven clinical decision-making.”