Table 2.

Predictability of hospitalizations per predictive model.

Model	F1 optimization, mean (SD)			Recall-oriented (Fβ=2) optimization, mean (SD)
	F1-score	Recall	F1-score		Recall
KNNa	0.544 (0.007)	0.883 (0.020)	0.545 (0.005)		0.890 (0.017)
DTb	0.562 (0.030)	0.837 (0.069)	0.548 (0.004)		0.897 (0.007)
RFc	0.535 (0.010)	0.878 (0.016)	0.541 (0.005)		0.874 (0.029)
XGBd	0.546 (0.004)	0.897 (0.012)e	0.545 (0.004)		0.895 (0.011)
LRf	0.582 (0.006)e	0.878 (0.015)	0.583 (0.010)e		0.879 (0.015)
MLPg	0.549 (0.006)	0.892 (0.010)	0.551 (0.003)		0.897 (0.011)e
LGBMh	0.545 (0.005)	0.893 (0.013)	0.545 (0.005)		0.893 (0.016)

^aKNN: k-nearest neighbors.

^bDT: decision tree.

^cRF: random forest.

^dXGB: XGBoost.

^eBest-performing models

^fLR: logistic regression.

^gMLP: multilayer perceptron.

^hLGBM: LightGBM.