Demystifying machine learning in endourology - understanding models, applications, and clinical impact: a review from EAU endourology.

Abstract

Purpose of review: Machine learning algorithms are occupying a larger space in medical and urology applications. However, typical medical physicians are not trained on these technologies and do not master the possibilities offered by these tools, to imagine their applications in the medical field. This manuscript is indented to be a guide in the use of machine learning in different urology applications, and to demystify the available machine learning and artificial intelligence algorithms. This manuscript reviews some of their applications and potential applications to the medical and urology field.

Recent findings: Multiple works are published on the use of machine learning in urology, with performance demonstrated to be noninferior to human experts on multiple occasions. However, the major part of the machine learning publications in urology applications are concentrated on diagnosis and/or prognosis. Advanced machine learning algorithms based on agentic artificial intelligence, able to perform decisions and causality-based treatment optimization, are rarely put to use in urology. The democratization of advanced machine learning technologies in the medical fields can accelerate the adoption of these techniques, and potentially improve the patient care through relevant suggestive decision making.

Summary: This work aims to demystify the machine learning tools for medical applications, facilitate decision making and adoption of the correct tools for the correct applications, and places a roadmap for the future of machine learning in the enhancement of patient care in urology.