Skip to main content

Table 3 Predictive performance of each mapping algorithm

From: Mapping EORTC QLQ-C30 and FACT-G onto EQ-5D-5L index for patients with cancer

  Whole sample ninefold cross-validation
  RMSE MAE ρ RMSE MAE ρ
EORTC QLQ-C30
 Linear 0.099 0.075 0.838 0.100 0.076 0.833
 Beta 0.103 0.081 0.825 0.105 0.081 0.817
 Tweedie 0.110 0.084 0.803 0.114 0.086 0.799
 Tobit 0.102 0.077 0.836 0.103 0.078 0.822
 Two-part linear 0.100 0.075 0.837 0.101 0.076 0.825
 Two-part beta 0.099 0.075 0.840 0.101 0.077 0.828
 Ordinal logistic 0.100 0.077 0.835 0.101 0.078 0.829
FACT-G
 Linear 0.121 0.090 0.753 0.121 0.091 0.744
 Beta 0.121 0.091 0.754 0.122 0.092 0.752
 Tweedie 0.124 0.092 0.740 0.124 0.093 0.740
 Tobit 0.123 0.090 0.754 0.124 0.091 0.751
 Two-part linear 0.122 0.091 0.749 0.123 0.092 0.748
 Two-part beta 0.119 0.090 0.760 0.121 0.091 0.759
 Ordinal logistic 0.119 0.090 0.760 0.120 0.091 0.764
  1. The best performances in each performance measure in each source measure are italics
  2. RMSE, root mean squared error; MAE, mean absolute error; ρ, correlation coefficient; EORTC QLQ-C30, European Organization for Research and Treatment of Cancer Quality of Life Questionnaire Core 30; FACT-G, Functional Assessment of Cancer Therapy General