Artificial intelligence in genitourinary oncology: A bibliometric study and systematic review.

e16616 Background: Within the field of medicine, artificial intelligence (AI) is used to optimize diagnostic capacity, treatment planning, and prognostic evaluation. Research on AI has advanced significantly, but a comprehensive evaluation of the utility of AI in genitourinary (GU) oncology remains underexplored. We sought to conduct a bibliometric study of existing literature and a systematic review of randomized controlled trials (RCTs) to describe the state of the science regarding the use and applications of AI in GU oncology. Methods: We searched MEDLINE (Ovid), Embase (Ovid), and CINAHL Ultimate databases using search terms relevant to the concepts of GU oncology and AI. We excluded non-English papers, non-human studies, review articles, and articles using AI in manuscript writing. Our bibliometric study described articles from 2013-2023 using the term AI, and we categorized manuscripts by study type and cancer type. We also conducted a systematic review of RCTs assessing the use of AI in GU oncology. We used Covidence for screening and data extraction. Two authors independently reviewed all papers and Cochrane Risk of Bias (RoB) Tool 2.0 was used to assess for bias in the RCT studies. Results: The initial search identified 2,409 articles. After abstract review, 1,220 articles remained: 962 retrospective articles, 175 prospective studies, 79 with combined retrospective/prospective methods, and 4 RCTs. We also categorized studies by cancer type: 923 prostate, 274 renal, 194 urothelial, 8 testicular, and 2 penile cancers. AI-related articles grew exponentially from 14 in 2013 to 362 in 2023, with substantial growth starting in 2019 (92 that year). For our systematic review, we identified 4 RCTs: 1 in bladder cancer (BCa) and 3 in prostate cancer (PCa). Among the 4 RCTs, 2 focused on AI-based diagnostics, and the other 2 analyzed AI’s role in prognosis prediction and treatment planning. Of the diagnostic studies, 1 demonstrated that a neural network analyzing urinary biomarkers outperformed traditional methods in BCa diagnosis. The other demonstrated AI-enhanced imaging’s superior efficiency in detecting PCa. The third article demonstrated AI’s prognostic value in automated bone scan index for evaluating bone metastasis in PCa. The fourth study showed improved operational efficacy of AI-generated treatment plans compared to conventional brachytherapy planning in PCa. The 4 RCTs had varying levels of risk of bias, primarily due to the randomization process and deviations from intended interventions. Conclusions: Our bibliometric analysis of AI in GU oncology demonstrates the growing recognition of AI’s potential to enhance cancer care. Our systematic review identified four RCTs that highlight the diverse applications of AI and showcase AI’s ability in diagnostics and treatment planning. Collectively, this work provides information about the promise of AI in oncology while improving clinical outcomes.