Skip to main content
. 2021 Feb 18;18(8):1739–1745. doi: 10.7150/ijms.51235

Table 3.

Predictive performance of machine learning algorithms

Algorithms Mortality ICU
AUC (SD) Sensitivity Specificity AUC (SD) Sensitivity Specificity
Random Forests 89.0% (1.3%) 76.4% 89.5% 78.1% (3.1%) 73.4% 79.6%
Xgboost 88.4% (1.9%) 30.3% 96.8% 78.9% (2.9%) 54.2% 86.0%
Kernel SVM 87.8% (4.3%) 20.8% 99.5% 76.1% (2.2%) 43.3% 92.3%
Neural network 84.4% (2.6%) 36.8% 90.3% 71.8% (4.4%) 56.7% 77.8%
Logistic 81.7% 26.2% 90.1% 66.9% 55.3% 78.5%