UroPredict: Machine learning model on real-world data for prediction of kidney cancer recurrence (UroCCR-120)

This study developed a machine learning (ML) model, UroPredict, to predict individual postoperative disease-free survival (DFS) in kidney cancer patients. Using the UroCCR database, a cohort of 3372 surgically treated RCC patients was analyzed. A Cox PH model with 24 variables achieved an AUC of 0.81 (95% CI 0.77–0.85) in predicting DFS, outperforming conventional risk scores. Patients were stratified into four prognostic groups with good discrimination (iAUC = 0.79). The ML model offers accurate individual DFS prediction, aiding in personalized follow-up and adjuvant therapy selection.

Gaëlle Margue, Loïc Ferrer, Guillaume Etchepare, Pierre Bigot, Karim Bensalah, Arnaud Mejean, Morgan Roupret, Nicolas Doumerc, Alexandre Ingels, Romain Boissière, Géraldine Pignot, Bastien Parrière, Philippe Paparel, Thibaut Waeckerl, Thierry Colin, Jean-Christophe Bernhard