Table 5.

Effects of data augmentation and dropout techniques applied to the NTUH_ROCF^a data set.

Metric	Delayed recall trial, mean (SD)
	Without data augmentation and dropout techniques	With data augmentation and dropout techniques
Sensitivity	0.824 (0.019)	0.847 (0.017)
Specificity	0.898 (0.024)	0.905 (0.009)
Accuracy	0.861 (0.011)	0.876 (0.010)
AUROCb	0.893 (0.012)	0.913 (0.004)

^aNTUH_ROCF: National Taiwan University Hospital_Rey-Osterrieth Complex Figure.

^bAUROC: area under the receiver operating characteristic curve.