Table 2.

Macroaveraged scores for the machine learning algorithms.

Models		Precision, %	Recall, %	F1 score, %	Accuracy, %
Data set 1: MIMIC-IIIa (intravenous push patients)
	Random forest	68.96	68.75	68.70	68.75
	Adaptive boosting	74.37	72.92	72.80	72.92
	Support vector machine	85.19	73.33	73.79	73.33
	Extreme gradient boosting	79.27	76.25	77.58	76.25
	Shallow neural network	88.05	86.67	87.26	86.67
Data set 2: MIMIC-III (intravenous drip patients)
	Random forest	66.71	65.33	65.06	65.33
	Adaptive boosting	77.29	77.33	77.30	77.33
	Support vector machine	84.59	71.33	71.71	71.33
	Extreme gradient boosting	77.45	77.33	77.38	77.33
	Shallow neural network	85.99	86.00	85.98	86.00
Data set 3: eICUb (intravenous drip patients)
	Random forest	66.77	66.56	65.59	68.06
	Adaptive boosting	78.03	77.78	77.65	77.78
	Support vector machine	84.74	76.39	76.19	76.39
	Extreme gradient boosting	79.16	79.17	78.85	79.17
	Shallow neural network	87.80	87.50	87.55	87.50

^aMultiparameter Intelligent Monitoring In Intensive Care III database.

^beICU: e–Intensive Care Unit database.