From: An interpretable risk prediction model for healthcare with pattern attention
Sepsis prediction | Mortality prediction | |||||||
---|---|---|---|---|---|---|---|---|
10 h | 8 h | 6 h | 4 h | 10 h | 8 h | 6 h | 4 h | |
LR | .713 ± .017 | .739 ± .013 | .744 ± .017 | .753 ± .018 | .803 ± .014 | .820 ± .017 | .843 ± .014 | .866 ± .017 |
RF | .726 ± .014 | .741 ± .016 | .759 ± .019 | .762 ± .016 | .813 ± .016 | .829 ± .020 | .856 ± .016 | .873 ± .017 |
SVM | .701 ± .021 | .754 ± .019 | .761 ± .021 | .763 ± .018 | .802 ± .021 | .835 ± .015 | .853 ± .020 | .863 ± .018 |
GRU | .748 ± .011 | .761 ± .009 | .766 ± .008 | .767 ± .009 | .836 ± .011 | .842 ± .009 | .861 ± .010 | .882 ± .010 |
LSTM | .750 ± .011 | .763 ± .010 | .767 ± .011 | .770 ± .011 | .857 ± .009 | .862 ± .011 | .873 ± .009 | .884 ± .011 |
RETAIN | .741 ± .011 | .756 ± .009 | .758 ± .008 | .759 ± .008 | .848 ± .007 | .860 ± .008 | .875 ± .009 | .895 ± .008 |
IFM | .746 ± .009 | .760 ± .010 | .763 ± .009 | .765 ± .009 | .866 ± .007 | .873 ± .008 | .882 ± .008 | .892 ± .007 |
\({\text {PAVE}}^{-V}\) | .754 ± .006 | .768 ± .005 | .769 ± .005 | .770 ± .006 | .886 ± .006 | .886 ± .005 | .897 ± .005 | .915 ± .006 |
\({\text {PAVE}}^{-T}\) | .756 ± .007 | .770 ± .006 | .770 ± .006 | .774 ± .006 | .889 ± .008 | .892 ± .008 | .903 ± .008 | .918 ± .007 |
PAVE | .763 ± .005 | .773 ± .006 | .778 ± .004 | .780 ± .004 | .897 ± .005 | .904 ± .004 | .913 ± .005 | .922 ± .005 |