This study introduces a novel AI-driven grading system for prostate cancer (PCa), addressing the limitations of the current Gleason score system. Using multi-institutional datasets from 2647 PCa patients, the AI model developed a four-risk-group system superior to the existing five-grade groups, significantly improving biochemical recurrence and cancer-specific death prediction. The system also predicts castration-resistant PCa risk in men with biochemical recurrence, demonstrating interpretability and potential for clinical integration.

Okyaz Eminaga, Ferd Saad, Zhe Tian, Ulrich Wolfgang, Pierre I. Karakiewicz, Véronique Ouellet, Feryel Azzi, Tilmann Spieker, Burkhard M. Helmke, Markus Graefen, Xiaoyi Jiang, Lei Xing, Jörn H. Witt, Dominique Trudel, Sami-Ramzi Leyh-Bannurah