Table 2.
Diagnostic accuracy of diabetes based on IR spectra of saliva
| n | Accuracy (%) | SP (%)a | PPV (%)b | NPV (%)c | ||
|---|---|---|---|---|---|---|
| Training Set | ||||||
| Diabetes | 28 | 0 | 100.0 | 100.0 | 100.0 | 100.0 |
| Non-diabetes | 0 | 8 | 100.0 | 100.0 | 100.0 | 100.0 |
| Validation Set | ||||||
| Diabetes | 9 | 0 | 100.0 | 75.0 | 81.8 | 100.0 |
| Non-diabetes | 2 | 6 | 75.0 | 100.0 | 100.0 | 81.8 |
Overall accuracy: 100% on the training set, 88.2% on the test set. Diagnosis of diabetes was determined by linear discriminant analysis of the infrared spectra. Bold numbers are indicative of accurate classifications. Underlined numbers are indicative of inaccurate classifications. The accuracy column also represents sensitivity for diabetic patients (but not control subjects). SP = specificity; PPV = positive predictive value; NPV = negative predictive value.