This prognostic study created optimized ensembles of calibrated random forest models to predict clinically significant prostate cancer (csPCa, grade group ≥2 PCa) using total prostate-specific antigen (PSA), free PSA, negative biopsy status, and age, with or without DRE and MRI data. Observational data were aggregated from cohorts in six organizations in Canada, the USA, and Czechia. Prostate biopsies were performed between 2009 and 2024. Risk models (ClarityDX Prostate + DRE, ClarityDX Prostate + MRI, and ClarityDX Prostate + MRI + DRE) were derived (training cohorts n = 1626 to 2191) and validated (validation cohorts n = 378 to 1318) from different clinical sites. The models had ROC AUC values ≥ 0.80. Adding DRE improved the ROC AUC to 0.82 while models using MRI features had ROC AUC values of 0.87 (without DRE) and 0.88 (with DRE) in the validation cohort. These four ClarityDX Prostate models offer high accuracy in predicting csPCa in individuals in variable clinical settings.
NPJ digital medicine. 2026 Apr 17 [Epub ahead of print]
Robert J Paproski, Adam Kinnaird, M Eric Hyndman, Adrian Fairey, Leonard Marks, Christian P Pavlovich, Sean A Fletcher, Roman Zachoval, Vanda Adamcova, Jiri Stejskal, Armen Aprikian, Christopher J D Wallis, Desmond Pink, Catalina Vasquez, Perrin H Beatty, John D Lewis
Nanostics Inc., Edmonton, AB, Canada., Division of Urology, Department of Surgery, University of Alberta, Kipnes Urology Centre, Edmonton, AB, Canada., UCLA Health, Los Angeles, CA, USA., James Buchanan Brady Urological Institute, Johns Hopkins University School of Medicine, Baltimore, MD, USA., Department of Urology, 3rd Faculty of Medicine of Charles University and Thomayer University Hospital, Prague, Czech Republic., Division of Urology, Department of Surgery, University of Toronto, Toronto, ON, Canada., Nanostics Inc., Edmonton, AB, Canada. .