Data used for validation | Training data* | Validation data* | ||||||
---|---|---|---|---|---|---|---|---|
Sensitivity, % | PPV, % | Sensitivity, % | PPV, % | |||||
DC | C | DC | C | DC | C | DC | C | |
Fold 0 | 90 | 100 | 93 | 100 | 89 | 89 | 88 | 89 |
Fold 1 | 85 | 100 | 94 | 100 | 86 | 92 | 94 | 93 |
Fold 2 | 90 | 100 | 94 | 100 | 87 | 91 | 90 | 91 |
Fold 3 | 90 | 100 | 94 | 100 | 88 | 91 | 92 | 92 |
Fold 4 | 89 | 100 | 94 | 100 | 91 | 94 | 91 | 89 |
Note: C = concept model, DC = document–concept model, PPV = positive predictive value.
↵* Folds 0 and 1: patients in the training (n = 1945) and validation (n = 487) cohorts; and folds 2, 3 and 4: patients in the training (n = 1946) and validation (n = 486) cohorts.