Table 1 Machine learning algorithms applied in salivary spectra to discriminate between vehicle and CHIKV mice.
Pre-Processing | Algorithm | Accuracy | Sensitivity | Specificity |
|---|---|---|---|---|
Raw data (3050–2800 1800—900 cm-1) | Linear discriminant analysis | 0.48 | 0.33 | 0.62 |
Support vector machine | 0.51 | 0.5 | 0.52 | |
Amide I norm (3050–2800 1800—900 cm-1) | Linear discriminant analysis | 0.51 | 0.5 | 0.52 |
Support vector machine | 0.49 | 0.33 | 0.62 | |
Savitzky-Golay (3050–2800 1800—900 cm-1) | Linear discriminant analysis | 0.72 | 0.56 | 0.86 |
Support Vector Machine | 0.85 | 0.83 | 0.86 | |
Poly + Amide I (3050–2800 1800—900 cm-1) | Linear Discriminant Analysis | 0.59 | 0.5 | 0.67 |
Support Vector Machine | 0.67 | 0.5 | 0.66 | |
Als + Amide I (3050–2800 1800—900 cm-1) | Linear Discriminant Analysis | 0.79 | 0.72 | 0.86 |
Support Vector Machine | 0.72 | 0.61 | 0.81 | |
Rubberband + Amide 1 (3050–2800 1800—900 cm-1) | Linear Discriminant Analysis | 0.48 | 0.55 | 0.43 |
Support Vector Machine | 0.56 | 0.55 | 0.57 |
