Table 1.
Prediction results from the neural network on 3,344 RDTs from 10 brands with respect to human reading
Model | TP | TN | FP | FN | Sensitivity, % | Specificity, % |
Alco | 5 | 19 | 0 | 0 | 100.0 | 100.0 |
Avioq | 207 | 104 | 1 | 5 | 97.6 | 99.0 |
Biolidics | 143 | 190 | 0 | 3 | 97.9 | 100.0 |
Biotime | 18 | 7 | 0 | 0 | 100.0 | 100.0 |
Biosynex | 63 | 126 | 0 | 1 | 98.4 | 100.0 |
NGBiotech Cassette | 628 | 621 | 1 | 4 | 99.4 | 99.8 |
NGBiotech All in one | 310 | 538 | 3 | 1 | 99.7 | 99.4 |
Nova | 199 | 67 | 0 | 4 | 98.0 | 100.0 |
Realy | 16 | 11 | 0 | 0 | 100.0 | 100.0 |
Solo Lab | 16 | 9 | 0 | 0 | 100.0 | 100.0 |
Vedalab | 18 | 6 | 0 | 0 | 100.0 | 100.0 |
Global | 1,623 | 1,698 | 5 | 18 | 98.9 | 99.7 |
The true positives (TPs), true negatives (TNs), false positives (FPs), false negatives (FNs), sensitivity, and specificity are reported per device and for the whole set.