Table 5 Results of the reader study to validate the clinical utility of the deep learning model.
From: A novel deep learning model for a computed tomography diagnosis of coronary plaque erosion
Sensitivity (%) (95% CI) | Specificity (%) (95% CI) | PPV (95% CI) | NPV (95% CI) | FPR (%) (95% CI) | FNR (%) (95% CI) | |
|---|---|---|---|---|---|---|
Novel DL model | 87.1 (70.2–96.4) | 85.3 (75.3–92.4) | 71.1 (58.3–81.2) | 94.1 (86.5–97.6) | 14.7 (7.6–24.7) | 12.9 (3.6–29.8) |
Before DL model assistance | ||||||
Reader 1 | 16.1 (5.5–33.7) | 77.3 (66.2–86.2) | 22.7 (10.6–42.1) | 69.1 (64.7–73.1) | 22.7 (13.8–33.8) | 83.9 (66.3–94.5) |
Reader 2 | 12.9 (3.6–29.8) | 88.0 (78.4–94.4) | 30.8 (12.9–57.2) | 71.0 (67.6–74.1) | 12.0 (5.6–21.6) | 87.1 (70.2–96.4) |
Reader 3 | 16.1 (5.5–33.7) | 76.0 (64.8–85.1) | 21.7 (10.2–40.5) | 68.7 (64.2–72.8) | 24.0 (14.9–35.2) | 83.9 (66.3–94.5) |
After DL model assistance | ||||||
Reader 1 | 83.9 (66.3–94.6) | 89.3 (80.1–95.3) | 76.5 (62.4–86.4) | 93.1 (85.7–96.8) | 10.7 (4.7–9.9) | 16.1 (5.4–33.7) |
Reader 2 | 77.4 (58.9–90.4) | 85.3 (75.3–92.4) | 68.6 (55.0–79.6) | 90.1 (82.6–94.6) | 14.7 (7.6–24.7) | 22.6 (9.6–41.1) |
Reader 3 | 77.4 (58.9–90.4) | 85.3 (75.3–92.4) | 68.6 (55.0–79.6) | 90.1 (82.6–94.6) | 14.7 (7.6–24.7) | 22.6 (9.6–41.1) |
