Table 1 Results of the validation of algorithms
Algorithm used | Algorithm identified | Validated diagnosis of epilepsy (n = 100) | Validated diagnosis of no epilepsy (n = 100) | Positive predictive value (95%CI) | Sensitivity (95% CI) | Negative predictive value (95% CI) | Specificity (95%CI) | F Score |
|---|---|---|---|---|---|---|---|---|
Drugs (iteration 1) | Yes | 24 | 0 | 100% | 24% (16–33) | 75% (73–77) | 100% (96–100) | 0.39 |
No | 76 | 100 | ||||||
Keywords (iteration 2) | Yes | 86 | 6 | 86% (74–93) | 86% (78–92) | 94% (91–96) | 94% (87–98) | 0.86 |
No | 14 | 94 | ||||||
Both (iteration 3) | Yes | 87 | 7 | 84% (72–91) | 87% (79–93) | 94% (91–97) | 93% (86–97) | 0.856 |
No | 13 | 93 | ||||||
Final algorithm (iteration 4) | Yes | 95 | 1 | 98% (85–100) | 95% (89–98) | 98% (95–99) | 99% (95–100) | 0.96 |
No | 5 | 99 |