Table 1. Model performance for GAM models considering different prevalence levels.

Prevalence	AUC	Sensitivity	Specificity
10%	0.947±0.019	0.922±0.029	0.889±0.041
30%	0.944±0.017	0.926±0.037	0.872±0.041
50%	0.945±0.021	0.946±0.036	0.864±0.044
70%	0.943±0.019	0.930±0.027	0.875±0.044
90%	0.918±0.025	0.912±0.043	0.826±0.057
Median	0.944±0.003	0.953±0.011	0.843±0.001

Performance is shown in terms of classification rate (AUC, sensitivity and specificity). Sensitivity and specificity correspond to those calculated using the MST threshold; the validation set used here represents 20% of the data (the other 80% was used to calibrate the models).