Table 3.
Misclassified samples.
| S1 (XRT14) | S2 (XRT36) | S3 (XRT39) | S4 (XRT33) |
|---|---|---|---|
| 57 | 129 | 87 | 342 |
| 78 | 257 | 105 | 492 |
| 136 | 327 | 201 | 1354 |
| 122 | 309 | 183 | 1514 |
| 79 | 180 | 125 | 765 |
| 92 | 126 | 168 | 3 41 |
| 42 | 44 | 54 | 946 |
| 103 | 175 | 184 | 1045 |
| 41 | 34 | 49 | 1430 |
| 62 | 178 | 258 | 52 |
| 77 | 234 | 183 | 1142 |
| 97 | 286 | 374 | 82 |
| 146 | 239 | 232 | 1388 |
| 162 | 167 | 137 | 2518 |
Notes: Misclassified samples. Expressions for the 14 most discriminatory probes. Samples S1, S2 and S3 were predicted to be High Fatigue and S4 to be Low Fatigue. The expression values for S1, S2 and S3 were closer to the mean expression of the High Fatigue group in the learning phase. Conversely, the expression values for S4 is closer to the Low Fatigue group. S1, S2 and S3 might define a new group of Low Fatigue with very small expressions (lower than the corresponding expressions observed among High Fatigue subjects) in this reduced base of 14 genes.