Copyright: ©Author(s) 2026.
World J Clin Oncol. Nov 24, 2026; 17(11): 126582
Published online Nov 24, 2026. doi: 10.5306/wjco.126582
Published online Nov 24, 2026. doi: 10.5306/wjco.126582
Table 1 Characteristics of the analyzed development and second-institution cohorts, mean ± SD/n (%)
| Characteristic | Development cohort (n = 1035) | Second-institution cohort (n = 212) |
| Age, years | 47.78 ± 13.68 | 48.88 ± 13.57 |
| Sex | ||
| Female | 457 (44.15) | 90 (42.45) |
| Male | 578 (55.85) | 122 (57.55) |
| Location | ||
| Upper extremity | 150 (14.49) | 31 (14.62) |
| Lower extremity | 504 (48.70) | 95 (44.81) |
| Trunk | 381 (36.81) | 86 (40.57) |
| Depth | ||
| Superficial | 307 (29.66) | 60 (28.30) |
| Deep | 728 (70.34) | 152 (71.70) |
| Tumor size, cm | 5.95 ± 4.23 | 6.39 ± 4.12 |
| FNCLCC grade | ||
| 1 | 464 (44.83) | 92 (43.40) |
| 2 | 246 (23.77) | 48 (22.64) |
| 3 | 325 (31.40) | 72 (33.96) |
| Surgical margin | ||
| Negative | 952 (91.98) | 190 (89.62) |
| Positive | 83 (8.02) | 22 (10.38) |
| Radiotherapy | ||
| No | 658 (63.57) | 110 (51.89) |
| Yes | 377 (36.43) | 102 (48.11) |
| Chemotherapy | ||
| No | 793 (76.62) | 154 (72.64) |
| Yes | 242 (23.38) | 58 (27.36) |
| Bone or neurovascular invasion | ||
| No | 802 (77.49) | 163 (76.89) |
| Yes | 233 (22.51) | 49 (23.11) |
| Unplanned surgery | ||
| No | 822 (79.42) | 178 (83.96) |
| Yes | 213 (20.58) | 34 (16.04) |
| Symptom duration, months | 24.66 ± 13.21 | 24.86 ± 13.84 |
| Local recurrence within 3 years | 331 (32.0) | 67 (31.60) |
Table 2 Repeated nested internal-evaluation performance by modeling strategy, mean ± SD
| Strategy | AUC | Brier score | Calibration slope |
| CatBoost | 0.8126 ± 0.0052 | 0.1584 ± 0.0017 | 1.070 ± 0.080 |
| Equal-weight top-3 ensemble | 0.8134 ± 0.0043 | 0.1586 ± 0.0016 | 1.086 ± 0.044 |
| Brier-weighted top-3 ensemble | 0.8133 ± 0.0043 | 0.1586 ± 0.0016 | 1.086 ± 0.044 |
| Gradient boosting | 0.8144 ± 0.0031 | 0.1589 ± 0.0011 | 1.117 ± 0.019 |
| XGBoost | 0.8124 ± 0.0035 | 0.1595 ± 0.0011 | 1.054 ± 0.027 |
| Logistic top-3 stack | 0.8117 ± 0.0044 | 0.1608 ± 0.0023 | 0.962 ± 0.038 |
| Histogram gradient boosting | 0.8092 ± 0.0026 | 0.1613 ± 0.0012 | 0.912 ± 0.018 |
| Random forest | 0.8058 ± 0.0053 | 0.1620 ± 0.0024 | 1.134 ± 0.073 |
| RBF support vector machine | 0.8005 ± 0.0045 | 0.1634 ± 0.0022 | 0.976 ± 0.051 |
| Extra trees | 0.8015 ± 0.0056 | 0.1637 ± 0.0030 | 1.067 ± 0.090 |
| Elastic-net logistic | 0.7994 ± 0.0017 | 0.1640 ± 0.0009 | 1.042 ± 0.011 |
| LightGBM | 0.8023 ± 0.0048 | 0.1642 ± 0.0018 | 0.837 ± 0.035 |
| Multilayer perceptron | 0.7955 ± 0.0087 | 0.1647 ± 0.0026 | 0.867 ± 0.119 |
| Full 12-variable logistic | 0.7998 ± 0.0021 | 0.1647 ± 0.0011 | 0.907 ± 0.013 |
| Shrinkage LDA | 0.7974 ± 0.0021 | 0.1657 ± 0.0011 | 0.878 ± 0.015 |
| Six-variable logistic benchmark | 0.7948 ± 0.0013 | 0.1663 ± 0.0005 | 0.948 ± 0.009 |
| k-nearest neighbors | 0.7629 ± 0.0077 | 0.1802 ± 0.0018 | 1.012 ± 0.178 |
Table 3 Protocol-frozen second-institution performance of CatBoost and logistic comparators (95%CI)
| Model | AUC | Brier score | Log loss | Calibration intercept | Calibration slope | O/E ratio |
| CatBoost | 0.8050 (0.7433-0.8668) | 0.1615 (0.1331-0.1912) | 0.4884 (0.4146-0.5678) | 0.148 (-0.222 to 0.490) | 0.985 (0.703-1.357) | 1.079 (0.885-1.276) |
| Six-variable logistic | 0.8134 (0.7533-0.8734) | 0.1572 (0.1310-0.1866) | 0.4770 (0.4116-0.5509) | -0.044 (-0.408 to 0.305) | 0.978 (0.710-1.333) | 0.979 (0.811-1.151) |
| Full 12-variable logistic | 0.8104 (0.7488-0.8719) | 0.1580 (0.1297-0.1863) | 0.4800 (0.4066-0.5549) | -0.059 (-0.432 to 0.272) | 0.921 (0.691-1.259) | 0.973 (0.806-1.129) |
- Citation: Li ZF, Yang BC, Liu YB, Liu SQ, Zhang Q, Yu SJ. Machine learning and logistic regression for ascertainable three-year recurrence status after soft tissue sarcoma surgery. World J Clin Oncol 2026; 17(11): 126582
- URL: https://www.wjgnet.com/2218-4333/full/v17/i11/126582.htm
- DOI: https://dx.doi.org/10.5306/wjco.126582