Most of the available evidence derives from studies in prostate cancer, where multimodal AI models, most prominently the ArteraAI platform,2 have demonstrated robust performance in stratifying patients beyond conventional clinical risk groups. Notably, these models were able to identify subsets of patients undergoing radiotherapy who derive benefit from androgen deprivation therapy (ADT), thereby enabling potential treatment de-escalation in biomarker-negative individuals. This represents a clinically meaningful step toward reducing overtreatment while maintaining oncologic safety.
In bladder cancer, DP-AI models demonstrated a promising ability to predict response to intravesical therapies in non-muscle-invasive disease. AI-driven biomarkers identified patients unlikely to respond to Bacillus Calmette-Guérin (BCG), who may instead benefit from alternative regimens such as gemcitabine/docetaxel.3 These findings are of high clinical relevance, especially in the context of ongoing BCG shortages and the need for improved treatment allocation.
Beyond predictive applications, the prognostic performance of DP-AI models was consistently strong across disease stages and tumour types. In renal cell carcinoma, AI-based pathomics signatures achieved high accuracy in predicting recurrence and survival outcomes. In contrast, evidence in testicular and penile cancers remains scarce, primarily reflecting the rarity of these malignancies and the resulting limited availability of large, well-annotated datasets required for robust AI model development and validation.
Despite these promising findings, several limitations must be acknowledged. Most studies included were retrospective, often based on post hoc analyses or relatively small cohorts, raising concerns regarding selection bias and overfitting. Importantly, none of the identified DP-AI biomarkers have yet been validated in prospective settings.
The integration of DP-AI into clinical workflows holds considerable promise. These models can be seamlessly incorporated into routine histopathological assessment, potentially enhancing diagnostic precision, improving prognostic accuracy, and supporting personalized treatment decisions. Moreover, AI-driven approaches may contribute to increased efficiency and reduced healthcare costs by streamlining workflows and minimizing unnecessary treatments.
DP-AI biomarkers represent a rapidly evolving and highly promising tool in urologic oncology. While current evidence supports their prognostic and, in selected cases, predictive utility, prospective validation remains essential before widespread clinical implementation. Future research should focus on large-scale, multi-institutional prospective studies and the integration of AI with other emerging modalities, such as advanced imaging and genomic profiling, to fully realize the potential of personalized cancer care.
Written by:
- Navid Roessler, Department of Urology, Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria; Department of Urology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany
- Shahrokh F. Shariat, Professor and Chairman, Department of Urology, Comprehensive Cancer Center, Medical University, Vienna, Vienna General Hospital, Währinger Gürtel, Vienna, Austria
- Roessler N, Miszczyk M, Miyajima K, Dematteis A, Alfarhan AR, Cormio A, Alqahtani AS, Fazekas T, Schuettfort VM, Vetterlein MW et al: Harnessing Artificial Intelligence for Risk Stratification and Outcome Prediction in Urologic Cancers: A Systematic Review. Eur Urol Focus 2025.
- Esteva A, Feng J, van der Wal D, Huang SC, Simko JP, DeVries S, Chen E, Schaeffer EM, Morgan TM, Sun YL et al: Prostate cancer therapy personalization via multi-modal deep learning on randomized phase III clinical trials. Npj Digital Medicine 2022, 5(1).
- Packiam VT, McElree IM, Ghodoussipour S, Nimgaonkar V, Krishna V, Kim JK, Allison DB, Richards JR, Anand Rajan KD, Chen SJ et al: Presence of an Artificial Intelligence-powered Predictive Biomarker Is Associated with a Poor Response to Intravesical Bacillus Calmette-Guerin but Not to Intravesical Sequential Gemcitabine/Docetaxel in Patients with High-grade Non-muscle-invasive Bladder Cancer. Eur Urol Oncol 2025.