Table 6 Performance on the NTUH dataset.

From: An Automatic Assessment System for Alzheimer’s Disease Based on Speech Using Feature Sequence Generator and Recurrent Neural Network

 

Manual

Automatic Transcription

N/A

Without Fine-Tuning

After Fine-Tuning

thr = 3, biGRU (AUROC)

0.808 ± 0.05

0.803 ± 0.03

0.838 ± 0.03

thr = 3, biGRU (Sensitivity)

0.736 ± 0.07

0.711 ± 0.05

0.756 ± 0.07

thr = 3, biGRU (Specificity)

0.750 ± 0.05

0.822 ± 0.03

0.764 ± 0.06

Back to article page