Skip to main content
. 2022 Jun 11;10(6):1087. doi: 10.3390/healthcare10061087

Table 6.

Comparisons between different models constructed using 12 h dataset in terms of their prediction accuracy.

Metric Method In-Hospital Mortality Short-Term Mortality Long-Term Mortality
48 h 72 h 30 Days 1 Year
AUROC CART 0.8101 0.8033 0.8006 0.7925 0.8471
LR 0.8029 0.8659 0.8222 0.8224 0.9082
MARS 0.8124 0.8502 0.8195 0.8170 0.8716
RF 0.8415 0.8867 0.8498 0.8543 0.8953
GB 0.8489 0.8862 0.8542 0.8556 0.9171
Specificity CART 0.7416 0.8425 0.7071 0.7544 0.8181
LR 0.7352 0.7921 0.7455 0.7743 0.8521
MARS 0.5696 0.6243 0.5861 0.6344 0.7090
RF 0.1735 0.2408 0.1638 0.1900 0.3796
GB 0.7528 0.8712 0.7578 0.7907 0.9259
Sensitivity CART 0.7460 0.7211 0.7864 0.7553 0.8080
LR 0.7126 0.8188 0.7457 0.7158 0.8000
MARS 0.8599 0.8842 0.8665 0.8275 0.8480
RF 0.9912 0.9842 0.9970 0.9933 0.9760
GB 0.7810 0.7211 0.7864 0.7647 0.6720
Precision CART 0.2323 0.1167 0.2179 0.2605 0.0788
LR 0.2273 0.1041 0.2400 0.2756 0.0952
MARS 0.1731 0.0636 0.1785 0.2059 0.0531
RF 0.1117 0.0361 0.1101 0.1231 0.0294
GB 0.2487 0.1391 0.2520 0.2950 0.1487
F1-Statistic CART 0.3542 0.2009 0.3413 0.3874 0.1436
LR 0.3447 0.1848 0.3632 0.3979 0.1702
MARS 0.2882 0.1187 0.2960 0.3297 0.1000
RF 0.2007 0.0696 0.1983 0.2191 0.0571
GB 0.3773 0.2332 0.3817 0.4258 0.2435

CART, classification and regression trees; LR, logistic regression; MARS, multivariate adaptive regression splines; RF, random forest; and GB, gradient boosting.