Table 3 Statistical results of MLP-1, MLP-2, and MLP-3 for their independent test datasets (n = 3). Accuracy, specificity × 100, and sensitivity × 100 reported with 95% confidence interval radius (CIR).

From: Machine Learning and Feature Selection Applied to SEER Data to Reliably Assess Thyroid Cancer Prognosis

 

MLP-1

MLP-2

MLP-3

Accuracy ± 95% CIR (%)

94.49 ± 0.88

91.09 ± 0.71

80.87 ± 1.71

Alive (Specificity × 100) ± 95% CIR (%)

94.45 ± 0.90

91.08 ± 0.72

80.84 ± 1.75

COD-TC (Sensitivity × 100) ± 95% CIR (%)

96.36 ± 4.95

91.41 ± 4.86

81.40 ± 8.22

Threshold

0.0447

0.028

0.0319

MCC

0.501

0.383

0.304

PPV

0.277

0.180

0.158

NPV

0.999

0.998

0.990

F1 Score

0.431

0.301

0.265

Back to article page