1109 Artificial Intelligence in Urological Oncology. (12th October 2021)

• Record Type: Journal Article
• Authors: Brodie, A; Dai, N; Teoh, J Y C; Decaestecker, K; Dasgupta, P; Vasdev, N
• Is Part Of: British journal of surgery. Volume 108:Supplement 6(2021)
• Publication Date: 2021-10-12
• Subjects: Surgery -- Periodicals; 617.005
• DOI: 10.1093/bjs/znab259.947
• Languages: English
• ISSNs: 0007-1323

Abstract

Abstract: Aim: A comprehensive review of the literature on the current and future applications of artificial intelligence (AI) in the context of urological oncology. Method: Four key areas of urological oncology were identified, and a comprehensive literature review was carried out in each area looking at the current and future applications of AI. These four areas included: Prostate cancer, Renal cancer, Bladder cancer, Robotic Surgery. Results: In total, 63 primary research articles were reviewed across these four areas. For prostate, renal and bladder cancer, AI has already shown great promise in the areas of imaging and histopathology interpretation, predicting tumour grade, reducing inter-observer variability and identification of genomic biomarkers. For robotic surgery, AI has already demonstrated value in the assessment of operator skill and using this to predict surgical outcomes. However, some common limitations to the applicability of AI into clinical practice include an overwhelming predominance of small retrospective studies, concerns over the datasets and methodology of AI training, the complexity of AI algorithms being such that they become un-interpretable and the technological requirements and ethical considerations with so much confidential "big data." Conclusions: The potential for AI to improve clinical care is clearly unparalleled but there remain significant challenges to adoption into clinical practice. Future research will need to focus on the establishment of multi-institute open access databases and improved data collection and integration for improved training of AI algorithms and ultimately, for clinical applicability to be realised, there needs to be high-quality prospective randomised multi-institute studies.