Table 4.

Performance metrics across datasets with a model trained on the ADNI training set, using a cut-off of and Alzheimer’s disease (AD) score of 0.5 and employing inference using MC Dropout with 50 samples.

Dataset	Accuracy	AUC	Sensitivity	Specificity	PPV/precision	NPV
ADNI test set	0.92	0.97	0.90	0.93	0.90	0.93
NACC (only AD/Control)	0.74	0.79	0.68	0.78	0.65	0.80
NACC (AD/All)	0.72	0.76	0.68	0.73	0.56	0.83

AUC area under the ROC curve, PPV positive predictive value, NPV negative predictive value.