Table 1.

The medians and 95% confidence intervals for all adopted metrics are reported and out-of-bag (OOB) performances of models with and without spatial cross-validation (CV) are compared.

Metric (%)	OOB Performance	Spatial CV
Acc	72.9 (72.1, 73.3)	69.7 (63.6, 75.7)
F1	65.1 (64.3, 65.7)	62.0 (53.2, 69.0)
Sens	82.5 (81.8, 83.6)	78.7 (69.9, 87.8)
Prec	53.7 (52.8, 54.5)	50.9 (43.0, 59.1)
Spec	68.7 (67.6, 69.2)	69.2 (48.9, 76.0)
AUC	84.1 (83.8, 84.5)	81.3 (76.0, 84.8)

We evaluated our machine learning models in terms of accuracy (Acc), F1 score, sensitivity (Sens), precision (Prec), specificity (Spec) and area under the ROC curve (AUC).