Table 2 Training and test set results for the 5-year recurrence prediction model.

From: Recurrence prediction in clear cell renal cell carcinoma using machine learning of quantitative nuclear features

 

Prediction

Rec ( +)

Rec ( −)

Total

(a) Training set result

 Accuracy: 92.7%

  Truth

   Rec ( +)

708

166

874

   Rec ( −)

77

2378

2455

   Total

785

2544

3329

(b) Test set result (ROI-based)

 Accuracy: 86.4%

  Truth

   Rec ( +)

222

53

275

   Rec ( −)

81

627

708

   Total

303

680

983

(c) Test set result (case-based)

 Accuracy: 100%

  Truth

   Rec ( +)

10

0

10

   Rec ( −)

0

21

21

   Total

10

21

31

Back to article page