Table 1.

Error Probability using a Least Squares Linear Classifier (LSLC) for the best number of features as function of the window length.

Combination of Signals	Par.	Window Length
		10 s	20 s	40 s	60 s
ECG	Error(%)	43.0%	41.2%	40.1%	39.6%
	$N_{max}$	174	80	174	80
	p-value	<0.001	<0.001	<0.001	Best
TEB	Error(%)	51.0%	42.2%	34.6%	37.8%
	$N_{max}$	60	60	40	20
	p-value	<0.001	<0.001	Best	<0.001
ECG+TEB+EDA	Error(%)	26.6%	27.9%	22.2%	24.1%
	$N_{max}$	20	80	40	20
	p-value	<0.001	<0.001	Best	<0.001
ECG+TEB	Error(%)	41.9%	31.3%	25.7%	27.1%
	$N_{max}$	80	80	60	40
	p-value	<0.001	<0.001	Best	<0.001
ECG+EDA	Error(%)	26.0%	28.3%	27.9%	29.2%
	$N_{max}$	40	40	40	10
	p-value	Best	<0.001	<0.001	<0.001
TEB+EDA	Error(%)	29.9%	31.2%	29.7%	30.9%
	$N_{max}$	20	20	40	20
	p-value	<0.001	<0.001	Best	<0.001
EDA	Error(%)	36.1%	37.3%	36.5%	37.1%
	$N_{max}$	20	20	20	20
	p-value	Best	0.003	<0.001	<0.001