Table 7.
Results of end-to-end evaluation and the corresponding 95% CIs for each change class
| Results | Accuracy (CI) (%) | Precision (CI) (%) | Recall (CI) (%) | F1 score (CI) (%) |
|---|---|---|---|---|
| New or worse | 99.3 (98.2–100.0) |
92.6 (87.6–97.6) |
92.6 (87.6–97.6) |
92.6 (87.6–97.6) |
| Unchanged | 98.9 (97.3–100.0) |
93.1 (88.2–98.0) |
96.4 (92.9–99.9) |
94.7 (90.4–99.0) |
| Improved | 99.8 (99.4–100.0) |
100.0 (99.7–100.0) |
85.7 (78.9–92.5) |
92.3 (87.2–97.4) |
| Indeterminate | 98.2 (95.9–100.0) |
99.4 (98.4–100.0) |
98.5 (96.5–100.0) |
99.0 (97.5–100.0) |