Table 2 Assessment of logistic regression and random forest biomarker models in the confirmatory and replication cohorts

From: Tandem mass tag-based quantitative proteomic profiling identifies candidate serum biomarkers of drug-induced liver injury in humans

Method

Biomarkers/Models

AUC of confirmatory cohort between NDO vs. DO

AUC of replication cohort between NDO vs. DO

Logistic regression

FBP1 + GSTA1

0.75

0.69

FBP1 + GSTA1 + LECT2

0.78

0.68

FBP1 + CES1 + LECT2

0.78

0.64

Random forest

FBP1 + LECT2

1.00

0.64

FBP1 + LECT2 + CPS1

1.00

0.61

  1. AUC area under the receiver operating characteristic curve, NDO non-DILI patients at onset, DO DILI patients at onset. See also Supplementary Fig. 5.
Back to article page