Performance comparisons for unfavorable-outcome prediction (mRS >2) between the DLPD and the clinical readersa
| AUC | Sensitivity | Specificity | |
|---|---|---|---|
| Model/readers | |||
| Neuroradiologist I | 0.76 (0.64–0.85) | 0.68 (0.53–0.81) | 0.69 (0.54–0.83) |
| Neuroradiologist II | 0.81 (0.71–0.89) | 0.61 (0.47–0.77) | 0.86 (0.73–0.97) |
| Neuroradiology fellow | 0.82 (0.72–0.89) | 0.75 (0.62–0.87) | 0.69 (0.53–0.83) |
| Clinical readers | |||
| Neurologist I | 0.79 (0.69–0.88) | 0.82 (0.7–0.93) | 0.53 (0.36–0.68) |
| Neurologist II | 0.77 (0.67–0.86) | 0.64 (0.49–0.78) | 0.83 (0.7–0.95) |
| Consensus read | 0.79 (0.68–0.87) | 0.70 (0.56–0.83) | 0.75 (0.60–0.88) |
| DLPD | 0.81 (0.72–0.89) | 0.68 (0.54–0.81) | 0.81 (0.67–0.92) |
| P value | P = .005 | P = .25 | P = .03 |
↵a The data in the parentheses represent the 95% confidence interval. The P value is for the noninferiority test between the consensus clinical read and the DLPD (predefined margin, .05).