Table 5.

Predictability of survivability per predictive model.

Model		F1 optimization, mean (SD)		Fβ=2 (recall-oriented) optimization, mean (SD)
		F1-score	Recall	F1-score	Recall
At the time of hospitalization
	KNNa	0.616 (0.035)	0.735 (0.048)	0.546 (0.021)	0.901 (0.030)
	DTb	0.707 (0.013)	0.864 (0.011)	0.673 (0.022)	0.908 (0.017)
	RFc	0.696 (0.030)	0.901 (0.030)	0.666 (0.021)	0.666 (0.021)
	XGBd	0.765 (0.025)	0.834 (0.042)	0.726 (0.008)	0.920 (0.022)
	LRe	0.492 (0.012)	0.909 (0.019)	0.476 (0.035)	0.916 (0.022)
	MLPf	0.681 (0.024)	0.824 (0.027)	0.569 (0.020)	0.922 (0.023)
	LGBMg	0.761 (0.017)	0.874 (0.016)	0.717 (0.036)	0.922 (0.021)
At the time of intensive care unit admission
	KNN	0.582 (0.040)	0.740 (0.053)	0.527 (0.030)	0.885 (0.049)
	DT	0.652 (0.045)	0.879 (0.035)	0.638 (0.032)	0.922 (0.023)
	RF	0.630 (0.018)	0.908 (0.039)	0.587 (0.035)	0.941 (0.016)
	XGB	0.703 (0.035)	0.838 (0.068)	0.672 (0.021)	0.918 (0.051)
	LR	0.497 (0.018)	0.908 (0.031)	0.470 (0.049)	0.920 (0.028)
	MLP	0.633 (0.044)	0.790 (0.094)	0.529 (0.019)	0.935 (0.020)
	LGBM	0.701 (0.025)	0.886 (0.034)	0.672 (0.024)	0.915 (0.027)

^aKNN: k-nearest neighbors.

^bDT: decision tree.

^cRF: random forest.

^dXGB: XGBoost.

^eLR: logistic regression.

^fMLP: multilayer perceptron.

^gLGBM: LightGBM.