Table 1.
Results of Classification Modelsa
| A) | IMS-MS Screening Model Results | ||||||
|---|---|---|---|---|---|---|---|
| Passive Sampler Model | Pregnancy Model | ||||||
| Training | Testing | Training | Testing | ||||
| # Samples | 42 | 13 | # Samples | 87 | 28 | ||
| Accuracy (%) | 100 | 100 | Accuracy (%) | 90.8 | 92.9 | ||
| Sensitivity (%) | 100 | 100 | Sensitivity (%) | 92.7 | 94.4 | ||
| Specificity (%) | 100 | 100 | Specificity (%) | 87.1 | 90.0 | ||
| Kappa | 1 | 1 | Kappa | 0.80 | 0.84 | ||
| B) | Annotated Model Results | ||||||
| Passive Sampler Model | Pregnancy Model | ||||||
| Training | Testing | Training | Testing | ||||
| # Samples | 42 | 13 | # Samples | 87 | 28 | ||
| Accuracy (%) | 93.1 | 100 | Accuracy (%) | 72.4 | 67.9 | ||
| Sensitivity (%) | 86.7 | 100 | Sensitivity (%) | 85.5 | 77.8 | ||
| Specificity (%) | 96.7 | 100 | Specificity (%) | 49.5 | 50.0 | ||
| Kappa | 82.2 | 1 | Kappa | 0.37 | 0.28 | ||
a
(A) shows results using our IMS-MS screening method for passive sampler (left) and pregnancy data (right), while (B) shows results using annotated feature data. Metrics for the training data were calculated using 5-fold cross-validation.