Table 3 Prediction accuracy of 4- and 6-marker models

From: Proteomic analysis of lymphoblastoid cell lines from schizophrenic patients

Variable

1st sample set (CON n = 30, SCZ n = 30)

2nd sample set (CON n = 30, SCZ n = 30)

Combined (CON n = 60, SCZ n = 60)

OR

P-valuea

Prediction accuracyb

OR

P-value

Prediction accuracy

OR

P-value

Prediction accuracy

4-marker model

MX1

11.91

0.017

 

26.57

0.016

 

19.46

<0.001

 

GLRX3

0.06

0.036

81.7%

4.99

0.507

73.3%

0.46

0.384

77.5%

UROD

0.11

0.011

(AUC = 0.86)

0.55

0.609

(AUC = 0.72)

0.44

0.108

(AUC = 0.82)

GART

0.02

0.006

 

0.01

0.006

 

0.01

<0.001

 

6-marker model

MX1

7.68

0.079

 

16.72

0.049

 

20.48

<0.001

 

GLRX3

0.04

0.03

 

2.37

0.741

 

0.48

0.431

 

UROD

0.07

0.012

81.7%

0.79

0.875

78.3%

0.51

0.235

77.5%

GART

0.01

0.011

(AUC = 0.88)

0.003

0.008

(AUC = 0.66)

0.01

<0.001

(AUC = 0.82)

MAPRE1

0.19

0.095

 

103.8

0.052

 

0.73

0.629

 

TBCB

0.13

0.050

 

1.93

0.616

 

0.42

0.138

 
  1. OR odds ratio, Combined combined results of 1st sample set and 2nd sample set, AUC the area under the receiver operating characteristic curves
  2. aP-value: multivariate logistic regression analysis
  3. bPrediction accuracy: [1—overall misclassification rate (OMR)] × 100 (%)
Back to article page