Effect of Missing Data Imputation on Deep Learning Prediction Performance for Vesicoureteral Reflux and Recurrent Urinary Tract Infection Clinical Study
Table 3
10-fold cross-validation results on CNTK.
Accuracy
SD
Sensitivity
SD
Specificity
SD
Deep learning-original dataset
57.65
4.18
58.09
4.09
57.32
7.32
FAMD-ncomp = 2
58.55
3.58
58.99
3.99
59.52
5.62
FAMD-ncomp = 3
61.52
1.08
60.20
0.99
61.00
1.32
FAMD-ncomp = 6
58.85
0.10
55.89
0.16
65.92
0.26
FAMD-ncomp = 10
57.5
0.14
54.4
0.20
63.90
0.29
MICE
64.05
4.38
64.59
0.09
62.62
6.12
ncomp: number of component; SD: standard deviation.