Current state of AI for shockwave lithotripsy: a systematic review from YAU and EAU endourology.

PURPOSE

To consolidate the current evidence of artificial intelligence (AI) for management of nephrolithiasis using extracorporeal shock-wave lithotripsy (ESWL), and to look at its feasibility into integration in clinical practice.

METHODS

In March 2025, the sytematic search was conducted across several databases, including, PubMed, Google Scholar, ACM Digital Library, CINAHL, and IEEE Xplore via Boolean operators with the use of the dedicated terms. Studies that described the development and validation and/or testing of AI-based models for ESWL, regardless of clinical scenario, published in English were included.

RESULTS

17 studies met the inclusion criteria. These were grouped based on their primary application into two key scenarios: outcomes prediction (n = 14) and intraoperative assistance (n = 3). Despite promising results, many studies used meaningfully different methodologies to develop AI-based models, especially different baseline inputs. Moreover, many studies present mutually exclusive information, as illustrated by the use of body mass index (BMI) as input. Finally, many studies are presented as single center studies or without external testing, which reduces the likelihood of generalizability of the resulting accuracy metrics.

CONCLUSION

There is increasing evidence of the role of AI in predicting ESWL outcomes and assisting during the procedure, often outperforming traditional statistical models. More prospective multi-institutional studies need to be done with standardized parameters and external validation to fully integrate AI in the management of ESWL.