Table 8.
Quantification of the human-in-the-loop audit for AI-generated cases.
| Quality control metric | Occurrences/time per case, mean (SD) |
| Medical indicators/threshold errors corrected | 1.65 (0.68) |
| Logical inconsistencies/unreasonable plots rectified | 2.43 (1.08) |
| Terminology/expression normative revisions | 3.12 (1.12) |
| Average expert review and audit time (minutes) | 18.72 (2.89) |