Table 1. Improvements in prediction performances when new markers, , are added to a baseline model (), respectively.
Performance Measure | ||||
AUC | Gini | Pietra | sBrier | |
Model | ||||
B | 0.822 | 0.644 | 0.485 | 0.306 |
B+M 1 | 0.841 | 0.683 | 0.521 | 0.344 |
B+M 2 | 0.844 | 0.687 | 0.568 | 0.363 |
Absolute (Relative) Improvement | ||||
from B to B+M 1 | +0.019 (+2.3%) | +0.039 (+6.1%) | +0.036 (+7.4%) | +0.038 (+12.4%) |
from B to B+M 2 | +0.022 (+2.7%) | +0.043 (+6.7%) | +0.083 (+17.1%) | +0.057 (+18.6%) |
The discrimination power of is independent of the baseline score, whereas that of is concentrated in the gray zone of the baseline model.