Meta-analysis of 14 studies of AI for CT urinary stone detection/classification: pooled sensitivity 0.967, specificity 0.979; detection studies lower (0.935, 0.919). Most had high risk of bias and no external validation, so routine use is not supported.
Meta-analysisn = 14 studies
Talyshinskii et al. Too good to be true? AI performance in CT-based urinary stone detection and classification: EAU endourology systematic review and diagnostic accuracy meta-analysis