Table 2 This table summarizes the models developed for the prediction of SDM, their inputs and performance
From: Navigating advanced renal cell carcinoma in the era of artificial intelligence
| Â | Model | Input | Performance | Limitations |
|---|---|---|---|---|
Bai et al. | MRI radiomics-based nomogram for the prediction of SDM | Radiomic-score and SDM-related clinic-radiologic characteristics in 201 patients | -Training: 0.914 -Internal validation: 0.854 -External validation: 0.816 | -Retrospective study -Some SDM not pathologically proven -normogram developed using 3T, contrast-enhanced MRI - No real-world validation |
Wen et al. | Radiomics model for preoperative prediction of SDM in ccRCC patients. | Quantitative extraction of shape, size and texture-based features in contrast-enhanced CT scan imaging of 172 subjects from The Cancer Imaging Archive (TCIA) | -Training: 0.890 -Internal validation: 0.830 -External validation: - | - Retrospective study - No external validation cohort - No real-world validation |
Yu et al. | Radiomics model for the prediction of SDM in ccRCC | Contrast-enhanced CT scan imaging & clinicopathologic data in 242 patients | -Training: 0.882 -Internal validation: 0.916 -External validation: 0.925 | - Retrospective study - Imbalance between study cohorts -CT acquisition parameters inconsistent -Some SDM not pathologically proven - No real-world validation |