From: Using ordinal outcomes to construct and select biomarker combinations for single-level prediction
Class | Model | μ=−1 | μ=0 | μ=1 | μ=2 | μ=3 |
---|---|---|---|---|---|---|
P(D=1) = 0.1 | ||||||
Binary | Simple | 0.976 (0.974, 0.978) | 0.920 (0.915, 0.924) | 0.773 (0.764, 0.780) | 0.530 (0.508, 0.543) | 0.670 (0.642, 0.684) |
 | Sequential | 0.974 (0.971, 0.976) | 0.920 (0.915, 0.924) | 0.773 (0.764, 0.780) | 0.532 (0.519, 0.544) | 0.720 (0.708, 0.729) |
Nominal | BaselineCat | 0.976 (0.974, 0.978) | 0.920 (0.915, 0.924) | 0.773 (0.764, 0.780) | 0.532 (0.519, 0.544) | 0.720 (0.707, 0.728) |
Ordinal | CumLogit | 0.970 (0.946, 0.975) | 0.918 (0.912, 0.923) | 0.776 (0.769, 0.783) | 0.544 (0.536, 0.552) | 0.313 (0.306, 0.320) |
 | AdjCatLogit | 0.970 (0.952, 0.975) | 0.918 (0.912, 0.923) | 0.776 (0.769, 0.783) | 0.544 (0.536, 0.552) | 0.313 (0.306, 0.320) |
 | ContRatLogit | 0.971 (0.958, 0.976) | 0.918 (0.912, 0.923) | 0.776 (0.769, 0.783) | 0.544 (0.536, 0.552) | 0.313 (0.306, 0.320) |
 | Stereo | 0.976 (0.974, 0.978) | 0.920 (0.915, 0.924) | 0.776 (0.769, 0.783) | 0.535 (0.520, 0.547) | 0.724 (0.715, 0.732) |
P(D=1) = 0.5 | ||||||
Binary | Simple | 0.950 (0.946, 0.952) | 0.920 (0.915, 0.924) | 0.841 (0.834, 0.848) | 0.714 (0.705, 0.723) | 0.588 (0.571, 0.599) |
 | Sequential | 0.924 (0.911, 0.933) | 0.919 (0.915, 0.924) | 0.842 (0.834, 0.848) | 0.712 (0.701, 0.722) | 0.743 (0.733, 0.752) |
Nominal | BaselineCat | 0.950 (0.946, 0.952) | 0.920 (0.915, 0.924) | 0.841 (0.835, 0.848) | 0.712 (0.702, 0.722) | 0.743 (0.733, 0.752) |
Ordinal | CumLogit | 0.054 (0.050, 0.062) | 0.916 (0.907, 0.921) | 0.844 (0.838, 0.849) | 0.721 (0.715, 0.728) | 0.599 (0.593, 0.604) |
 | AdjCatLogit | 0.073 (0.054, 0.198) | 0.917 (0.911, 0.922) | 0.844 (0.838, 0.849) | 0.721 (0.715, 0.728) | 0.599 (0.593, 0.604) |
 | ContRatLogit | 0.094 (0.057, 0.409) | 0.917 (0.911, 0.922) | 0.844 (0.838, 0.849) | 0.721 (0.715, 0.728) | 0.599 (0.593, 0.604) |
 | Stereo | 0.950 (0.947, 0.953) | 0.920 (0.915, 0.924) | 0.844 (0.838, 0.849) | 0.718 (0.709, 0.725) | 0.749 (0.741, 0.756) |