Photo Credit: Zentangle

The following is a summary of “A novel predictive method for URS and laser lithotripsy using machine learning and explainable AI: results from the FLEXOR international database,” published in the May 2025 issue of World Journal of Urology by Nedbal et al. 

Researchers conducted a retrospective study to develop machine learning (ML) algorithms for predicting ureteroscopy (URS) outcomes, aiming to enhance diagnosis, treatment planning, personalized care, and clinical decision-making.  

They utilized the FLEXOR database, a large international multicentric cohort of 6,669 patients treated with URS for urolithiasis between 2015 and 2023. Preoperative and postoperative (PO) correlations were explored using 15 ML-trained algorithms. Outcomes included stone-free status (SFS) assessed at 3-month imaging follow-up, intraoperative complications (PCS bleeding, ureteric/PCS injury, need for PO drainage), and PO complications (fever, sepsis, need for reintervention). The ML techniques were applied for prediction, correlation, and logistic regression analysis. Explainable artificial intelligence (AI) was used to identify and highlight key features and their contributions to the output.  

The results showed that the Extra Tree Classifier achieved the highest accuracy of 81% in predicting SFS and PCS bleeding was negatively associated with ‘positive urine culture’ (-0.08), ‘tamsulosin’ (-0.08), ‘stone location’ (-0.10), ‘fibre optic scope’ (-0.19), ‘Moses Fibre’ (-0.09), and ‘TFL’ (-0.09), while it was positively correlated with ‘elevated creatinine’ (0.25), ‘fever’ (0.11), and ‘stone diameter’ (0.21). Both ‘PCS injury’ and ‘ureteric injury’ had moderate correlations with ‘elevated creatinine’ (0.11), ‘fever’ (0.10), and ‘lower pole stone’ (0.09). Use of ‘tamsulosin’ (0.23), presence of ‘multiple’ (0.25) or ‘lower pole’ (0.25) stones, ‘reusable scope’ (0.17), and ‘Moses Fibre’ (0.25) increased the risk of PO stent placement, whereas ‘digital scope’ (-0.13) and ‘TFL’ (-0.29) reduced it. Preoperative fever (0.10), ‘positive urine culture’ (0.16), and ‘stone diameter’ (0.10) were associated with PO fever and sepsis, the SFS was most influenced by ‘age’ (0.12), ‘preoperative fever’ (0.09), ‘multiple stones’ (0.15), ‘stone diameter’ (0.17), ‘Moses Fibre’ (0.15), and ‘TFL’ (-0.28).  

Investigators concluded that ML proved to be a helpful tool for accurately predicting outcomes by analyzing pre-existing datasets, with the model demonstrating strong performance in predicting outcomes and risks, thus providing a foundation for the development of accessible predictive models. 

Source: link.springer.com/article/10.1007/s00345-025-05551-2