Supporting Risk-Aware Use of Online Translation Tools in Delivering Mental Healthcare Services among Spanish-Speaking Populations

Table 1

Performance of RVM classifiers with lexical dispersion features.

RVM	Training data		Testing data
	AUC mean	SD	AUC	Accuracy	Sensitivity	Specificity
Dispersion rates (full 20 features)	0.625	0.06	0.649	0.667	0.578	0.753
Disp all 20 with min-max normalisation	0.547	0.04	0.654	0.643	0.566	0.718
Disp all 20 with L2 normalisation	0.573	0.07	0.594	0.560	0.518	0.600
Disp all 20 with Z-score normalisation	0.561	0.06	0.645	0.637	0.530	0.741
D4 (automatically optimised)	0.617	0.06	0.648	0.661	0.566	0.753
D4 with min-max	0.611	0.06	0.686	0.649	0.542	0.753
D4 with L2	0.571	0.07	0.595	0.560	0.518	0.600
D4 with Z-score	0.610	0.06	0.689	0.649	0.566	0.729