Table 6.

Sensitivity, specificity and accuracy in split verification data and test data by learned RNN.

	Sensitivity	Specificity	Accuracy
Validation
Whole data	0.563 ± 0.062	0.882 ± 0.0241	0.803 ± 0.0291
Undersampling	0.751 ± 0.0312	0.686 ± 0.0213	0.718 ± 0.0222
Test
Whole data	0.114 ± 0.092	0.899 ± 0.0221	0.862 ± 0.0201
Undersampling	0.471 ± 0.092	0.661 ± 0.0563	0.652 ± 0.0523

Sensitivity, specificity and accuracy were calculated for RNN predictions and PCR results from 5 split validation groups for the whole data and undersampling sets respectively. For the Test data, the mean value and standard deviation were calculated for the results of prediction by the RNN that independently learned with each verification group. Numbers on superscript show groups in which no significant difference are detected. On the other combinations, significant differences were shown with the Student's and Welch’s t-tests.