Table 3 Mean performance using the standard grading system, logistic regression (LR), and Artificial Neural Network (ANN) to predict surgical resectability in patients with glioblastoma.

From: Improved Prediction of Surgical Resectability in Patients with Glioblastoma using an Artificial Neural Network

Parameter

Estimate (95% CI)

P value

Accuracy

ANN

83.4 (81.6–85.1)

<0.01

LR

83.2 (81.3–85.0)

<0.01

Standard grading system

80.2 (78.2–82.1)

Reference

AUC

ANN

0.871 (0.849–0.895)

<0.01

LR

0.868 (0.848–0.889)

<0.01

Standard grading system

0.786 (0.747–0.825)

Reference

Sensitivity

ANN

0.586 (0.531–0.640)

0.0861

LR

0.587 (0.532–0.643)

0.0807

Standard grading system

0.543 (0.489–0.560)

Reference

Specificity

ANN

0.915 (0.898–0.932)

0.0117

LR

0.910 (0.893–0.927)

0.0380

Standard grading system

0.885 (0.861–0.908)

Reference

PPV

ANN

0.707 (0.649–0.766)

0.0164

LR

0.679 (0.619–0.739)

0.236

Standard grading system

0.642 (0.585–0.698)

Reference

NPV

ANN

0.877 (0.862–0.892)

0.0945

LR

0.878 (0.862–0.893)

0.0677

Standard grading system

0.864 (0.849–0.880)

Reference

Back to article page