Skip to main content

Table 3 Accuracy of the regression model according to the TG-ROC and uncertainty interval methods

From: Developing a clinical decision tool based on electroretinogram to monitor the risk of severe mental illness

Two-Graph ROC

Uncertainty Interval

 

Cutoff points

0.36 & 0.76

Observed

  

Cutoff points

0.41 & 0.73

Observed

 

SMI

Control

Total

SMI

Control

Total

Predicted

Most likely, SMI

>  0.76

31(a)

2(b)

33

Predicted

Most likely, SMI

>0.73

33(a)

3(b)

36

Uncertain

[0.36–0.76]

26

21

47

Uncertain

[0.41–0.73]

23

15

38

Most likely, not SMI

< 0.36

3(c)

17(d)

20

Most likely, not SMI

< 0.41

4(c)

22(d)

26

Total

60

40

100

Total

60

40

100

Sensitivity = 0.91

Specificity = 0.89

Sensitivity = 0.89

Specificity = 0.88

Accuracy = 0.90

   

Accuracy = 0.89

   
  1. Note. Accuracy of the regression model predicting severe mental illness (SMI) with ERG measures, allowing for an uncertainty level defined according to the two-graph ROC and uncertainty interval methods on the testing data (n = 100)
  2. Cutoff points represent two probability values of the ERG regression model that define the three levels of certainty. Sensitivity = a/(a + c). Specificity = d/(b + d). Accuracy: (a + d)/(a + b + c + d)