Attention: Restrictions on use of AUA, AUAER, and UCF content in third party applications, including artificial intelligence technologies, such as large language models and generative AI.
You are prohibited from using or uploading content you accessed through this website into external applications, bots, software, or websites, including those using artificial intelligence technologies and infrastructure, including deep learning, machine learning and large language models and generative AI.

ARTIFICIAL INTELLIGENCE From Diagnosis to Discharge: Artificial Intelligence in the Surgical Management of Urologic Patients

By: Jeffrey Wei, MD, Fox Chase Cancer Center, Philadelphia, Pennsylvania; Michael Eppler, BA, Keck School of Medicine, University of Southern California, Los Angeles; Randall A. Lee, MD, Fox Chase Cancer Center, Philadelphia, Pennsylvania | Posted on: 19 Jan 2024

Artificial intelligence (AI) currently is in a state of rapid growth with widespread adoption across many fields. AI continues to become more ubiquitous within health care, though its utilization remains in its infancy. For urologists, AI represents a transformative opportunity for enhancing care at every stage of care for the surgical patient, including diagnosis, surgical management, and postoperative follow-up. AI-driven tools can streamline and enhance various aspects of patient management, from monitoring to decision-making. With the breadth of cases and distinctive diseases that urologists encounter, AI can be an invaluable tool for supporting patients and physicians.

In the preoperative stage, AI has utilization in both assisting physicians in diagnostic certainty as well as a patient resource. Currently, one of the most common applications of AI is for disease diagnosis. AI has demonstrated great promise in the diagnosis and detection of multiple urologic cancers.1,2 Within the realm of prostate cancer, one area in which AI models have already been deployed is addressing the variability between interpretation of multiparametric MRI for detection of prostate cancer.2 Performance of radiologists alone in detecting lesions concerning for prostate cancer had a pooled sensitivity of 79.5%. However, when augmented with AI, sensitivity increased to 89.1%, demonstrating superior performance in a partnership utilization model.2 By augmenting the performance of radiologic interpretations, imaging interpretation is just one of many areas where improved quality of diagnostic information available through AI will continue to improve preoperative patient selection and counseling.

On a more patient-facing note, the emergence of AI large language models (LLMs), such as ChatGPT, has the potential to augment the medical information available for patients. They can offer a user-friendly and human-like interaction, making medical information more accessible and understandable. These models have already shown they can answer clinical questions, working as a virtual resource and triage tool while delivering it in a human-like tone, potentially alleviating physician burden.3 However, popular LLMs such as ChatGPT are likely not ready for immediate adoption in health care. These models can occasionally provide responses that may be factually incorrect or overly generalized, and this poses a significant challenge in health care settings where precision and reliability are paramount.4 Despite these limitations, continued training and refinement of LLMs on specific health care data hold the promise of making these models reliable for counseling resources for patients. With careful development, oversight, and integration into urologic practice, AI can complement human health care providers and enhance the patient experience while ensuring the highest standards of care and accuracy.

One of the most exciting areas for urologists includes the utilization of AI within the operating room. In a high stakes environment where synthesis of multiple sources of data is critical, AI is already showing promise in enhancing performance and mitigating risks. During cystoscopy, AI computer vision can assist urologists by extracting microimaging structures and identifying pixel-level features undetectable by the human eye to improve rates of the diagnosis of bladder cancer intraoperatively.5 Where white light cystoscopy alone may leave clinically significant bladder cancer behind, AI can improve the chances of an accurate diagnosis.5 Recent utilization of the Cystoscopy Artificial Intelligence Diagnostic System showed higher accuracy (0.939, CI 0.902-0.964) and sensitivity (0.954, CI 0.902-0.983) in detecting carcinoma in situ and small bladder tumors with a shorter latency time than expert urologists. Beyond making urologists faster and more accurate in their interventions, utilization of AI offers multiple avenues for improvement in surgical performance. Video analysis by AI can already delineate steps of surgical procedures with reasonable accuracy and identify working instruments as well as track both visible and occluded structures.6 Current work out of the MUSIC consortium has shown how application of AI can be a powerful tool for assessing surgical performance for training and quality improvement.7 Even from a safety perspective, AI has also been developed for the automatic capture, detection, and analysis of adverse events.8 This convergence of AI and urology represents a paradigm shift in the field, promising improved patient outcomes and more efficient, safer, and skillful surgical practices.

After surgery, potential AI applications are wide ranging, from detection and prediction of cancer recurrence to optimizing perioperative management and even disseminating postoperative instructions to patients. In the postoperative period, AI tools like LLMs may soon offer roles in writing patient instructions upon hospital discharge.9 For the longer term, AI models naturally lend themselves to predicting patient outcomes as they can efficiently analyze and process vast amounts of patient data. By identifying intricate patterns and correlations from data stores, AI algorithms can generate predictive models which have already shown promise in predicting postoperative outcomes, and thus help urologists make data-driven treatment decisions. Recently, Huang et al showed a deep neural network to be a promising tool in predicting biochemical disease recurrence within 3 years of prostatectomy regardless of cancer grade group.10 For bladder cancer patients, artificial neural networks have been developed for predicting 5-year survival following radical cystectomy.11 By allowing urologists to better assess individual outcomes, patient treatment can become more comprehensive and individualized.

AI applications hold promise for various aspects of surgical care, including initial diagnosis, intraoperative adverse event identification, perioperative support, and even long-term prognosis. The types of AI discussed above are diverse, including neural networks, deep learning, and generative AI, each with their own potential to improve patient care and optimize outcomes. Urologists should become familiar with the increasing number and breadth of AI innovations that have potential within the perioperative period. The innovative nature of this technology necessitates further research to ensure its applicability and effectiveness across different scenarios. Additionally, extensive patient education will be essential, given possible potential reluctance to embrace AI in personalized health care.

  1. Xu X, Zhang X, Tian Q, et al. Three-dimensional texture features from intensity and high-order derivative maps for the discrimination between bladder tumors and wall tissues via MRI. Int J Comput Assist Radiol Surg. 2017;12(4):645-656.
  2. Ramacciotti LS, Hershenhouse JS, Moktar D, et al. Comprehensive assessment of MRI-based artificial intelligence frameworks performance in the detection, segmentation, and classification of prostate lesions using open-source databases. Urol Clin North Am. 2024;51(1):131-161.
  3. Lee P, Bubeck S, Petro J. Benefits, limits, and risks of GPT-4 as an AI chatbot for medicine. N Engl J Med. 2023;388(13):1233-1239.
  4. Davis R, Eppler M, Ayo-Ajibola O, et al. Evaluating the effectiveness of artificial intelligence–powered large language models application in disseminating appropriate and readable health information in urology. J Urol. 2023;210(4):688-694.
  5. Wu S, Chen X, Pan J, et al. An artificial intelligence system for the detection of bladder cancer via cystoscopy: a multicenter diagnostic study. J Natl Cancer Inst. 2022;114(2):220-227.
  6. Hashimoto DA, Rosman G, Witkowski ER, et al. Computer vision analysis of intraoperative video: automated recognition of operative steps in laparoscopic sleeve gastrectomy. Ann Surg. 2019;270(3):414-421.
  7. Ghani KR, Liu Y, Law H, et al. PD46-04 Video analysis of skill and technique (VAST): machine learning to assess surgeons performing robotic prostatectomy. J Urol. 2017;197(Suppl 4):e891.
  8. Eppler MB, Sayegh AS, Maas M, et al. Automated capture of intraoperative adverse events using artificial intelligence: a systematic review and meta-analysis. J Clin Med. 2023;12(4):1687.
  9. Patel SB, Lam K. ChatGPT: the future of discharge summaries?. Lancet Digit Health. 2023;5(3):e107-e108.
  10. Huang W, Randhawa R, Jain P, et al. A novel artificial intelligence–powered method for prediction of early recurrence of prostate cancer after prostatectomy and cancer drivers. JCO Clin Cancer Inform. 2022;6(6):e2100131.
  11. Lam KM, He X, Choi KS. Using artificial neural network to predict mortality of radical cystectomy for bladder cancer. 2014 International Conference on Smart Computing. 2014:201-207.