Table 1.
Patient and tumor characteristics of training set, validation set and test set.
| Characteristic | Total | Train | Val | Test | P value |
|---|---|---|---|---|---|
| No. | 3701 | 2222 | 736 | 743 | |
| Menopause status | 0.998 | ||||
| premenopausal | 1522 (41%) | 897 (40%) | 311 (42%) | 314 (42%) | |
| menopause | 2179 (59%) | 1325 (60%) | 425 (58%) | 429 (58%) | |
| Tumor size | 0.074 | ||||
| ≤2 cm | 1921 (52%) | 1147 (52%) | 368 (50%) | 406 (55%) | |
| å 2 cm | 1780 (48%) | 1075 (48%) | 368 (50%) | 337 (45%) | |
| Histological grade | 0.409 | ||||
| 1 | 227 (6%) | 153 (7%) | 34 (5%) | 40 (5%) | |
| 2 | 2616 (71%) | 1562 (70%) | 536 (73%) | 518 (70%) | |
| 3 | 858 (23%) | 507 (23%) | 166 (23%) | 185 (25%) | |
| Lymph node status | 0.052 | ||||
| negative | 1953 (53%) | 1173 (53%) | 387 (53%) | 393 (53%) | |
| ITCs | 118 (3%) | 67 (3%) | 29 (4%) | 22 (3%) | |
| micrometastasis | 564 (15%) | 340 (15%) | 95 (13%) | 129 (17%) | |
| macrometastasis | 1066 (29%) | 642 (29%) | 225 (31%) | 199 (27%) | |
| Tumor location | 0.558 | ||||
| UOQ | 2090 (56%) | 1251 (56%) | 427 (58%) | 412 (55%) | |
| UIQ | 865 (23%) | 511 (23%) | 176 (24%) | 178 (24%) | |
| LOQ | 441 (12%) | 276 (12%) | 74 (10%) | 91 (12%) | |
| LIQ | 305 (8%) | 184 (8%) | 59 (8%) | 62 (8%) | |
| Vasular invasion | 0.131 | ||||
| absent | 2975 (80%) | 1774 (80%) | 609 (83%) | 592 (80%) | |
| present | 726 (20%) | 448 (20%) | 127 (17%) | 151 (20%) | |
| Nerve invasion | 0.824 | ||||
| absent | 3199 (86%) | 1936 (87%) | 627 (85%) | 636 (86%) | |
| present | 502 (14%) | 286 (13%) | 109 (15%) | 107 (14%) | |
| ER | 0.874 | ||||
| negative | 781 (21%) | 466 (21%) | 158 (21%) | 157 (21%) | |
| positive | 2920 (79%) | 1756 (79%) | 578 (79%) | 586 (79%) | |
| PR | 0.600 | ||||
| negative | 957 (26%) | 584 (26%) | 190 (26%) | 183 (25%) | |
| positive | 2744 (74%) | 1638 (74%) | 546 (74%) | 560 (75%) | |
| HER2 | 0.965 | ||||
| negative | 2625 (71%) | 1577 (71%) | 522 (71%) | 526 (71%) | |
| positive | 1076 (29%) | 645 (29%) | 214 (29%) | 217 (29%) | |
| Ki67 | 0.234 | ||||
| <20% | 531 (14%) | 320 (14%) | 113 (15%) | 98 (13%) | |
| ≥20% | 3170 (86%) | 1902 (86%) | 623 (85%) | 645 (87%) | |
| TILs | 0.568 | ||||
| low | 2657 (72%) | 1572 (71%) | 543 (74%) | 542 (73%) | |
| middle | 768 (21%) | 481 (22%) | 145 (20%) | 142 (19%) | |
| high | 276 (7%) | 169 (8%) | 48 (7%) | 59 (8%) | |
| BI-RADS grade | 0.458 | ||||
| 4a | 1032 (28%) | 615 (28%) | 193 (26%) | 224 (30%) | |
| 4b | 679 (18%) | 431 (19%) | 119 (16%) | 129 (17%) | |
| 4c | 1052 (28%) | 617 (28%) | 224 (30%) | 211 (28%) | |
| 5 | 938 (25%) | 559 (25%) | 200 (27%) | 179 (24%) | |
| Molecular subtype | 0.451 | ||||
| Luminal | 2931(79%) | 1762(79%) | 580(79%) | 589(79%) | |
| HER2 over-expression | 580(16%) | 344(15%) | 115(16%) | 121(16%) | |
| TNBC | 190(5%) | 116(6%) | 41(5%) | 31(5%) |
P values < 0.05 were considered statistically significant, and all P values were two-sided. P value is obtained by comparing the validation set and testing set.
UOQ upper outer quadrant, UIQ upper inner quadrant, LOQ lower outer quadrant, LIQ lower inner quadrant.