Table 2.
Assessment metrics for the conservative versus surgical treatment groups on the testing set, including: F1, sensitivity (recall), precision (PPV), specificity, NPV and accuracy (from s.d.).
| Fl | Sensitivity (recall) | Precision (PPV) | Specificity | NPV | Accuracy | |
|---|---|---|---|---|---|---|
| Conservative treatment | ||||||
| Random forest | 0.559 (0.509–0.608) | 0.691 (0.619–0.764) | 0.472 (0.416–0.529) | 0.902 (0.882–0.923) | 0.959 (0.948–0.970) | 0.879 (0.864–0.895) |
| XGBoost | 0.548 (0.517–0.579) | 0.744 (0.667–0.820) | 0.438 (0.395–0.481) | 0.880 (0.859–0.902) | 0.965 (0.952–0.977) | 0.865 (0.853–0.877) |
| LightGBM | 0.546 (0.505–0.588) | 0.735 (0.631–0.839) | 0.442 (0.390–0.494) | 0.882 (0.846–0.917) | 0.964 (0.948–0.980) | 0.865 (0.841–0.890) |
| LR | 0.546 (0.470–0.622) | 0.462 (0.382–0.542) | 0.678 (0.568–0.788) | 0.972 (0.960–0.983) | 0.935 (0.919–0.950) | 0.914 (0.895–0.934) |
| Weight | 0.556 (0.523–0.590) | 0.736 (0.650–0.821) | 0.454 (0.403–0.505) | 0.887 (0.858–0.916) | 0.964 (0.951–0.977) | 0.870 (0.852–0.889) |
| Stack | 0.572 (0.528–0.617) | 0.785 (0.707–0.863) | 0.457 (0.396–0.517) | 0.880 (0.848–0.913) | 0.970 (0.957–0.983) | 0.869 (0.845–0.893) |
| Weight-stack | 0.568 (0.535–0.602) | 0.742 (0.648–0.837) | 0.467 (0.416–0.518) | 0.892 (0.865–0.919) | 0.965 (0.951–0.979) | 0.876 (0.860–0.891) |
| Surgical treatment | ||||||
| Random forest | 0.352 (0.322–0.382) | 0.537 (0.479–0.595) | 0.267 (0.226–0.307) | 0.862 (0.831–0.892) | 0.952 (0.940–0.964) | 0.834 (0.807–0.860) |
| XGBoost | 0.347 (0.303–0.392) | 0.614 (0.536–0.692) | 0.245 (0.200–0.289) | 0.824 (0.795–0.852) | 0.957 (0.944–0.971) | 0.805 (0.777–0.834) |
| LightGBM | 0.353 (0.288–0.419) | 0.550 (0.482–0.619) | 0.264 (0.203–0.325) | 0.850 (0.801–0.899) | 0.952 (0.939–0.965) | 0.824 (0.775–0.873) |
| LR | 0.189 (0.165–0.213) | 0.691 (0.593–0.789) | 0.111 (0.093–0.128) | 0.484 (0.398–0.570) | 0.942 (0.921–0.964) | 0.500 (0.425–0.576) |
| Weight | 0.353 (0.308–0.398) | 0.556 (0.532–0.580) | 0.260 (0.213–0.307) | 0.851 (0.822–0.880) | 0.954 (0.945–0.962) | 0.826 (0.797–0.856) |
| Stack | 0.377 (0.324–0.430) | 0.562 (0.506–0.619) | 0.287 (0.232–0.342) | 0.867 (0.834–0.901) | 0.954 (0.943–0.966) | 0.841 (0.807–0.875) |
| Weight-stack | 0.367 (0.326–0.408) | 0.586 (0.536–0.636) | 0.269 (0.226–0.311) | 0.851 (0.824–0.878) | 0.956 (0.944–0.968) | 0.828 (0.799–0.858) |