Concordance Between Device and Reference Standard: Diagnosis
| Reference: biopsy1or panel consensus2 | |||
|---|---|---|---|
| Device Decision | Benign | Malignant | All |
| Benign | 88/145 (60.7%) | 1/10 (10.0%) | 89/155 (57.4%) |
| Malignant | 57/145 (39.3%) | 9/10 (90.0%) | 66/155 (42.6%) |
| Specificity (95% CI) | 0.61 (0.53 - 0.68) | ||
| Sensitivity (95% CI)* | 0.90 (0.71 - 1.00) | ||
| NPV (95% CI) | 0.99 (0.93 - 0.99) | ||
| PPV (95% CI) | 0.14 (0.07 - 0.25) | ||
| Reference: panel consensus2 alone | |||
| Benign | 85/136 (62.5%) | 1/12 (8.3%) | 86/148 (58.1%) |
| Malignant | 51/136 (37.5%) | 11/12 (91.7%) | 62/148 (41.9%) |
| Specificity (95% CI) | 0.63 (0.54 - 0.70) | ||
| Sensitivity (95% CI)* | 0.92 (0.76 - 1.000) | ||
| NPV (95% CI) | 0.99 (0.93 - 0.99) | ||
| PPV (95% CI) | 0.18 (0.09 - 0.30) | ||
Notes: 1Biopsy results are used as reference standard in lieu of consensus dermatologist panel assessment when biopsy results are available. Biopsy results are available for 22 lesions. 2Panel consensus: only includes diagnosis where all panel members agreed, regardless of the number of members in the panel. Panel consensus was available for 148 lesions. 95% Confidence interval (CI) calculated accounting for the within-subject correlation using the Wilson method *Wilson method not estimable, Wald method used.