Table 3 Performance of the different models predicting the type of response to neoadjuvant chemotherapy
Model* | AUC (95% CI) | Accuracy (95% CI) | Se (95% CI) | Sp (95% CI) | PPV (95% CI) | NPV (95% CI) |
|---|---|---|---|---|---|---|
MRs (TRG1-2 vs. 3–5) | ||||||
1. All types of chemotherapy (n = 73) *radiomic/clinicopathologic | 0.876 (0.786 − 0.966) | 0.84 (0.73–0.91) | 83 (56–96) | 84 (73–91) | 50 (29–71) | 96 (88–99) |
NRs (TRG1-3 vs. 4–5) | ||||||
1. All types of chemotherapy (n = 77) *radiomic | 0.725 (0.610–0.839) | 0.70 (0.59–0.80) | 72 (56–84) | 68 (53–81) | 70 (55–82) | 70 (54–83) |
2. ECF/EOX + FLOT (n = 61) *radiomic | 0.760 (0.639–0.882) | 0.75 (0.63–0.86) | 74 (57–87) | 77 (60–89) | 77 (60–89) | 74 (57–87) |
3. Oxaliplatin-based (n = 53) *radiomic | 0.810 (0.692–0.928) | 0.77 (0.64–0.88) | 96 (85–100) | 56 (37–74) | 71 (55–84) | 93 (73–99) |
4. FLOT (n = 40) *radiomic | 0.907 (0.818 − 0.995) | 0.82 (0.67–0.93) | 73 (52–88) | 94 (77–99) | 94 (76–99) | 74 (54–88) |