Statistical confusion matrix of algorithm in our sample size and differences between the current study results and the previous algorithm results
| Study of any diabetic retinopathy | |||||
| TP | FP | TN | FN | Error α | Area under the curve (AUC) 0.9884 |
| 219 | 3 | 376 | 4 | 0.0079 | |
| Specificity (SP) | Sensitivity (S) | Positive predictive values (PPV) | Negative predictive values (NPV) | Error β | |
| 99.21% | 98.21% | 98.65% | 98.95% | 0.0179 | |
| Current study results | Original algorithm | ||||
| S | 98.21% (95% CI, 96.24 to 99.99) | 84% (95% CI, 83.46 to 84.49) | |||
| SP | 99.21% (95% CI, 97.70 to 99.99) | 88.5% (95% CI, 88.29 to 88.72) | |||
| PPV | 98.65% (95% CI, 96.24 to 99.98) | 63.8% (95% CI, 63.18 to 64.35) | |||
| NPV | 98.95% (95% CI, 83.46 to 84.49) | 95.8% (95% CI, 95.68 to 95.96) | |||
| Error α | 0.0079 | 0.115 | |||
| Error β | 0–0179 | 0.16 | |||
| AUC | 0.988 (95% CI, 0.978 to 0.99) | 0.876 (95% CI, 0.858 to 0.886) | |||
FN, false negatives; FP, false positives; TN, true negatives; TP, true positives.