Full-text inclusion screening on Review 2 (7741 abstracts)
87False Positives (FP)gpt-5
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| gpt-52025.12 | 87 | — | 20.94 | 98.75 | 92.04 | 20.94 | 0.7 | 0.21 | 0.06 | |
| gpt-5-nanovariant=base2025.12 | 267 | — | 14.72 | 96.17 | 89.13 | 26.65 | 0.64 | 0.3 | 0.35 | |
| gpt-5-nanovariant=asterisk2025.12 | 358 | — | 19.27 | 94.82 | 88.24 | 26.19 | 0.67 | 0.31 | 0.36 | |
| gpt-5-mini2025.12 | 405 | — | 44.61 | 94.2 | 89.92 | 42.06 | 0.65 | 0.26 | 0.16 | |
| gpt-4o-mini2025.12 | 1,213 | — | 26.4 | 82.62 | 77.77 | 12.55 | 0.59 | 0.59 | 0.68 |