Performance of machine learning models and the SPAN-100 index in 3 cohorts
| Cohorts/Models | Sensitivity (%) | Specificity (%) | Accuracy (%) | AUC |
|---|---|---|---|---|
| Full cohort | ||||
| Clinical features only (16 features) | 78.1 | 65.5 | 73.5 | 0.77 |
| Imaging features only (11 features) | 53.5 | 79.9 | 63.2 | 0.69 |
| Both clinical and imaging features (27 features) | 74.4 | 69.8 | 72.8 | 0.79 |
| Best-performing clinical and imaging features (6 features) | 72.2 | 74.0 | 72.8 | 0.80a |
| SPAN-100 | 80.6 | 64.3 | 73.5 | 0.78 |
| Recanalized | ||||
| Clinical features only (16 features) | 73.1 | 70.4 | 71.9 | 0.76 |
| Imaging features only (11 features) | 53.7 | 69.4 | 60.9 | 0.61 |
| Both clinical and imaging features (27 features) | 74.5 | 68.9 | 72.0 | 0.77 |
| Best-performing clinical and imaging features (6 features) | 76.9 | 69.9 | 73.8 | 0.79a |
| SPAN-100 | 78.8 | 63.8 | 71.9 | 0.76 |
| Nonrecanalized | ||||
| Clinical features only (16 features) | 80.0 | 65.8 | 74.1 | 0.78 |
| Imaging features only (11 features) | 63.3 | 67.8 | 64.3 | 0.70 |
| Both clinical and imaging features (27 features) | 71.3 | 80.5 | 73.3 | 0.81 |
| Best-performing clinical and imaging features (6 features) | 81.9 | 75.4 | 80.5 | 0.82a |
| SPAN-100 | 65.5 | 77.1 | 68.1 | 0.78 |
↵a Model with the highest AUC value.