Table 2.
Comparison of diagnostic accuracy of algorithm and non-algorithm groups for detection of seizure neonate
| Algorithm group (n=128) | Non-algorithm group (n=130) | Difference (95% CI*) | |
|---|---|---|---|
| Number of neonates with seizures | 32 | 38 | .. |
| Number of true positives† | 26 | 34 | .. |
| Number of true negatives‡ | 81 | 82 | .. |
| Number of false positives§ | 15 | 10 | .. |
| Number of false negatives¶ | 6 | 4 | .. |
| Sensitivity‖ (95% CI*) | 81·3% (66·7 to 93·3) | 89·5% (78·4 to 97·5) | −8·2% (−25·0 to 7·7) |
| Specificity** (95% CI*) | 84·4% (76·9 to 91·0) | 89·1% (82·5 to 94·7) | −4·8% (−14·1 to 4·6) |
| False detection rate†† (95% CI*) | 36·6% (22·7 to 52·1) | 22·7% (11·6 to 35·9) | 13·9% (−5·2 to 32·7) |
Bias-corrected.
Seizure neonate correctly classified as a seizure neonate by the clinical team.
Non-seizure neonate correctly classified as a non-seizure neonate by the clinical team.
Non-seizure neonate incorrectly classified as a seizure neonate by the clinical team.
Seizure neonate incorrectly classified as a non-seizure neonate by the clinical team.
Percentage of seizure neonates correctly classified (ie, true positives among total true positives and false negatives).
Percentage of non-seizure neonates correctly classified (ie, true negatives among total true negatives and false positives).
Percentage of neonates classified as seizure neonates by the clinical team who did not have seizures (ie, false positives among total true and false positives).