Fig. 5

Comparison between the PAIDM and pathologists. a ROC curve for the performance of the PAIDM versus six pathologists in identifying LGNMI. b ROC curve for the performance of the PAIDM versus six pathologists in identifying HGNMI. c ROC curve for the performance of the PAIDM versus six pathologists in identifying HGMI. d Diagnostic accuracy of the PAIDM versus six pathologists in classifying the LGNMI, HGNMI and HGMI subtypes. e The average accuracy of the PAIDM versus six pathologists in the classification task. Error bars represent the 95% confidence intervals. PAIDM  pathological artificial intelligence diagnostic model, ROC  receiver operating characteristic, AUC  area under the curve, LGNMI  low-grade non-muscle invasion, HGNMI  high-grade non-muscle invasion, HGMI  high-grade muscle invasion, JP  junior pathologist, IP  intermediate pathologist, SP  senior pathologist