Skip to main content

Table 3 Classification performances of various deep neural network architectures on Mini Mental Status Exam (MMSE) and Korean Longitudinal Study on Cognitive Aging and Dementia Neuropsychological Battery (KLOSCAD-N) indicated by the area under the receiver operator curve (AUC) via five-cross validation on train dataset

From: Deep learning based low-cost high-accuracy diagnostic framework for dementia using comprehensive neuropsychological assessment profiles

  

2D-CNN

2D-CNN Naïve

2D-CNN w/o SC

1D-CNN

1D-CNN w/o SC

FCN

FCN w/o SC

NasNet

MMSEa

mean

-

-

-

-

-

0.9702

0.9583

-

 

std

-

-

-

-

-

0.0144

0.0139

-

KLOSCAD-N

mean

0.9863

0.9850

0.9782

0.9848

0.9805

0.9830

0.9771

0.9813

 

std

0.0048

0.0058

0.0057

0.0053

0.0042

0.0060

0.0070

0.0046

  1. aSince MMSE is composed with only five dimension (four demographic variables and one MMSE total-score, the other architecture are not applicable except FCN
Back to article page