• Bladder cancer grading is crucial for treatment decisions, but the existing ISUP 2004 system is subjective, compromising its reliability and prognostic utility.
• By mathematically defining well-established WHO 2004 grading criteria as quantitative nuclear features (QNFs) and employing AI-driven image analysis to extract QNFs, we previously developed precise, reproducible models for expert consensus grading.
• QNFs serve as excellent building blocks for prognostic classifiers.
• Using QNFs, we externally validate grading models and create recurrence-free survival (RFS) models that outperform grades assigned by pathologists.

Minqi Xua¹, Katherine Lindalea¹, Ava Slotmana¹, Raquel Benitezb², Dan Winkowskic³, Robert J Goodinga¹, Amber Simpsona¹, Nuria Malatsb², David M Bermana¹.

1. Queen’s University, Kingston, Canada
2. Centro Nacional de Investigaciones Oncológicas , Madrid, Spain
3. Visiopharm A/S, Westminster, Colorado;