Table 1.

Accuracy and validation accuracy statistics (Kappa, sensitivity, and specificity) for the three models. See text for the conventional interpretation of the Kappa values. Models 2 and 3 have excellent performance on the training set compared to model 1. Model 1 performed poorly on the validation set compared to Models 2 and 3. Overall, Model 3 performed marginally better than Model 2.

	Kappa	Sensitivity	Specificity
Model 1
Accuracy statistics	0.54 ± 0.048	0.79 ± 0.033	0.74 ± 0.037
Validation accuracy statistics	0.20 ± 0.039	0.69 ± 0.094	0.66 ± 0.027
Model 2
Accuracy statistics	0.84 ± 0.036	0.93 ± 0.02	0.94 ± 0.014
Validation accuracy statistics	0.67 ± 0.063	0.87 ± 0.062	0.90 ± 0.011
Model 3
Accuracy statistics	0.84 ± 0.026	0.97 ± 0.013	0.96 ± 0.013
Validation accuracy statistics	0.64 ± 0.035	0.88 ± 0.069	0.91 ± 0.012