Table 4. Cross-validated PLS-DA modeling statistics for the prediction performance for NA-SERS typing of individual type 2V M. pneumoniae clinical isolates.
| Isolate | P1 Type | CV Sensitivity | CV Specificity | CV Class Error |
|---|---|---|---|---|
| 1: M129 | 1 | 1 | 0.971 | 0.01 |
| 2: FH | 2 | 0.8 | 0.943 | 0.13 |
| 3: WI11 | 2V | 1 | 0.98 | 0.01 |
| 1: M129 | 1 | 1 | 1 | 0 |
| 2: FH | 2 | 0.88 | 0.943 | 0.08 |
| 3: WI17 | 2V | 1 | 0.98 | 0.01 |
| 1: M129 | 1 | 1 | 0.971 | 0.01 |
| 2: FH | 2 | 0.84 | 1 | 0.08 |
| 3: CO12 | 2V | 1 | 1 | 0 |
| 1: M129 | 1 | 1 | 1 | 0 |
| 2: FH | 2 | 0.92 | 0.971 | 0.05 |
| 3: CO44 | 2V | 1 | 0.98 | 0.01 |
| 1: M129 | 1 | 1 | 1 | 0 |
| 2: FH | 2 | 1 | 0.943 | 0.03 |
| 3: SA18 | 2V | 1 | 1 | 0 |
| 1: M129 | 1 | 1 | 1 | 0 |
| 2: FH | 2 | 0.96 | 1 | 0.02 |
| 3: GA3 | 2V | 1 | 1 | 0 |
Three categories were incorporated to cross-validate the model: either type 1; type 2; or neither. Clinical isolates were treated as an unknown class and cross-validated sensitivity, specificity, and class error were based on their classification prediction score as neither type 1 or type 2 reference control strains (i.e. category 3).