Skip to main content

Table 3 Predictive algorithm outcomes using the random forest: confusion matrices

From: Protein oxidation, nitration and glycation biomarkers for early-stage diagnosis of osteoarthritis of the knee and typing and progression of arthritic disease

  Algorithm 1   Algorithm 2
Clinical class Predicted class Clinical class Predicted class
Control Disease eOA eRA Non-RA
Training set cross-validation
 Control 11 2 eOA 13 0 0
 Disease 0 33 eRA 0 6 4
  Non-RA 0 2 8
Test set cross-validation
 Control 33 4 eOA 30 0 0
 Disease 11 86 eRA 1 22 12
  Non-RA 0 5 27
Test set validation
 Control 2 35 eOA 18 2 10
 Disease 36 61 eRA 0 8 27
  Non-RA 1 0 31
  1. eOA early osteoarthritis, aOA advanced osteoarthritis, eRA early rheumatoid arthritis, aRA advanced rheumatoid arthritis, CML Nε-carboxymethyl-lysine, CEL Nε-(1- carboxyethyl)-lysine, MetSO methionine sulfoxide, DT dityrosine, FL Nε-fructosyl-lysine, MG-H1 methylglyoxal-derived hydroimidazolone, 3DG-H 3-deoxyglucosone-derived hydroimidazolone isomers, NFK N-formylkynurenine, 3-NT 3-nitrotyrosine, G-H1 glyoxal-derived hydroimidazolone, CMA Nω-carboxymethylarginine