Skip to main content

Table 2 Performance measures based on cross-validation (Multivariable Models)

From: A comparison of statistical methods for the detection of hepatocellular carcinoma based on serum biomarkers and clinical variables

Method (Model)1 LOOCV AUC (95% CI)2 LOOCV ACC3 3CV AUC (95% CI)4 3CV ACC (SD)5
stepPLR (λ = 0.1) (male only, with age) 0.95(0.93-0.98) 86.63 0.94(0.94-0.95) 86.03(4.78)
stepPLR (λ = 1) (male only, with age) 0.95(0.93-0.98) 87.16 0.95(0.95-0.96) 87.12(4.31)
stepPLR (λ = 10) (male only, with age) 0.94(0.92-0.97) 86.63 0.95(0.94-0.95) 87.49(4.31)
stepPLR (λ = 0.1) (male only, without age) 0.90(0.86-0.94) 80.35 0.90(0.89-0.90) 80.97(5.16)
stepPLR (λ = 1) (male only, without age) 0.89(0.84-0.93) 81.28 0.89(0.89-0.90) 80.71(5.09)
stepPLR (λ = 10) (male only, without age) 0.91(0.87-0.95) 83.96 0.91(0.90-0.91) 82.17(5.37)
stepPLR (λ = 0.1) (gender, with age) 0.95(0.92-0.97) 88.25 0.95(0.94-0.95) 87.96(3.20)
stepPLR (λ = 1) (gender, with age) 0.95(0.92-0.97) 88.25 0.95(0.94-0.95) 88.34(3.51)
stepPLR (λ = 10) (gender, with age) 0.94(0.92-0.97) 88.25 0.95(0.94-0.95) 88.31(3.39)
stepPLR (λ = 0.1) (gender, without age) 0.92(0.88-0.95) 84.84 0.92(0.91-0.92) 84.35(4.23)
stepPLR (λ = 1) (gender, without age) 0.92(0.89-0.95) 84.84 0.92(0.92-0.92) 84.59(4.09)
stepPLR (λ = 10) (gender, without age) 0.92(0.89-0.95) 83.71 0.92(0.91-0.92) 83.72(4.29)
mob(AAT|~.) with age, male only) 0.85(0.79-0.91) 83.43 0.81(0.80-0.82) 76.23(5.71)
mob(GP73|~.) with age, male only) 0.88(0.82-0.93) 83.98 0.87(0.87-0.88) 80.77(4.99)
mob(AFP|~.) with age, male only) 0.86(0.80-0.92) 81.22 0.87(0.87-0.88) 83.66(5.52)
mob(Kin|~.) with age, male only) 0.92(0.88-0.96) 86.19 0.86(0.85-0.86) 79.36(5.08)
mob(AAT|~.) without age, male only) 0.78(0.71-0.85) 75.93 0.82(0.81-0.83) 75.33(5.15)
mob(GP73|~.) without age, male only) 0.83(0.76-0.89) 76.47 0.86(0.85-0.86) 77.81(4.16)
mob(AFP|~.) without age, male only) 0.80(0.74-0.86) 72.72 0.83(0.82-0.84) 76.54(5.42)
mob(Kin|~.) without age, male only) 0.88(0.83-0.93) 85.03 0.86(0.85-0.87) 78.88(4.92)
mob(AAT|~.) with age, gender) 0.87(0.83-0.92) 82.57 0.86(0.78-0.94) 81.19(4.59)
mob(GP73|~.) with age, gender) 0.93(0.89-0.96) 87.50 0.91(0.90-0.91) 84.61(4.03)
mob(AFP|~.) with age, gender) 0.90(0.86-0.94) 83.33 0.89(0.88-0.89) 82.95(4.42)
mob(Kin|~.) with age, gender) 0.89(0.85-0.93) 83.33 0.89(0.88-0.89) 82.81(4.14)
mob(AAT|~.) without age, gender) 0.84(0.79-0.89) 84.46 0.87(0.86-0.87) 80.93(3.71)
mob(GP73|~.) without age, gender) 0.91(0.87-0.94) 83.33 0.89(0.88-0.89) 81.22(4.35)
mob(AFP|~.) without age, gender) 0.87(0.82-0.91) 82.95 0.87(0.87-0.87) 80.60(3.73)
mob(Kin|~.) without age, gender) 0.88(0.84-0.92) 84.85 0.89(0.89-0.90) 83.05(3.97)
  1. 1. Method utilized for analysis, stepPLR stepwise penalized logistic regression with or without age and/or gender, mob model-based CART, classification and regression trees developed with or without age and/or gender; 2) LOOCV AUC, leave one out cross validation area under the curve; 3) LOOCV ACC, leave one out cross validation prediction accuracy; 4) 3CV AUC, three fold cross validation area under the curve; 5) 3CV ACC, three fold cross validation prediction accuracy.