Class
|
Sensitivity
|
Specificity
|
PPV
|
NPV
|
---|
0
|
0.9
|
0.667
|
0.818
|
0.8
|
1
|
0.667
|
0.9
|
0.8
|
0.818
|
- Support vector machine predictor was built as a linear function of the log-ratios of the expressions of 8 genes screened by RT-PCR. According to our LOOCV results, the 8 selected genes were able to predict with an accuracy of 81% patients’ outcome with 90% sensitivity and 66% specificity.