Table 3.

Performance on the entire data set. The performance of various machine learning algorithms using the standard and embedding features on a data set combining data from both home and laboratory environments^a.

Algorithm	Standard features		Embedding features
	AUCb	Accuracy	AUC	Accuracy
SVMc	0.751	0.735	0.738	0.692
Random forest	0.745	0.720	0.726	0.708
LightGBM	0.753	0.720	0.737	0.693
XGBoost	0.750 d	0.741	0.722	0.689

^aModels using standard features perform better than the models using embedding features in terms of both binary accuracy and area under the curve. Although the performance of the models is almost similar in terms of area under the curve metric, XGBoost outperforms others by considering both the area under the curve and accuracy metrics simultaneously.

^bAUC: area under the curve.

^cSVM: support vector machine.

^dVariable outperforms all others by taking both area under the curve and accuracy into account.