Table 5.
Rates of misclassified cases for pairwise comparisons (training and validation dataset)
| Luminal A (n = 66) | Luminal B (n = 25) | TN (n = 41) | All others (n = 118) | |
|---|---|---|---|---|
| Luminal A (n = 66) | – |
Lum A vs. B: F, 9/8 (13.6/30.8%) MI, 12/13 (18.2/52%) |
– | – |
| Luminal B (n = 25) |
Lum A vs. B: F, 9/8 (13.6/30.8%) MI, 12/13 (18.2/52%) |
– |
Lum B vs. TN: F, 8/8 (32/19.5%) |
Lum B vs. others: F, 17/19 (68/16.1%) POE, 17/18 (68/15.3%) MI, 12/14 (48/11.9%) |
| TN (n = 41) | – |
Lum B vs. TN: F, 8/8 (32/19.5%) |
– | – |
| All others (n = 118) | – |
Lum B vs. others: F, 17/19 (68/16.1%) POE, 17/18 (68/15.3%) MI, 12/14 (48/11.9%) |
– | – |
F Fisher, MI mutual information, POE probability of error and average correlation, TN triple negative