Discrimination and calibration statistics from recalibrated predicted probability algorithms
36.1–37.5°C | ⩽36.0°C or >37.5°C | |||
CRIB II | CRIB II(−T) | CRIB II | CRIB II(−T) | |
Observed deaths, n (%) | 167 (7.3) | 150 (15.3) | ||
Predicted deaths, n (%) | 141.3 (6.2) | 164.8 (7.2) | 175.7 (17.9) | 152.2 (15.6) |
Brier score | 0.044 | 0.043 | 0.075 | 0.075 |
Spiegelhalter’s z-score | p = 0.02 | p = 0.4 | p = 0.9 | p = 0.7 |
AUC | 0.917 | 0.917 | 0.921 | 0.919 |
Calibration-in-the-small | 0.0005 | 0.0001 | 0.0018 | 0.0011 |
Calibration-in-the-large | 0.0001 | <0.0001 | 0.0007 | <0.0001 |
Cox’s regression: | ||||
Intercept α (SE) | 0.36 (0.15) | 0.00 (0.14) | −0.28 (0.13) | 0.00 (0.14) |
Slope β (SE) | 0.05 (0.06) | 0.98 (0.06) | 1.07 (0.08) | 1.04 (0.08) |
Test: α = 0, β = 1 | p = 0.02 | p = 0.9 | p = 0.01 | p = 0.9 |
Test: α = 0 | β = 1 | p = 0.008 | p = 0.8 | p = 0.004 | p = 0.8 |
Test: β = 1 | α | p = 0.4 | p = 0.8 | p = 0.4 | p = 0.6 |
AUC, area under the curve; CRIB, Clinical Risk Index for Babies; CRIB II(−T), CRIB II without temperature at admission.