Table 2.
Reader | Sensitivity lesion level (%) | Sensitivity patient level (%) | Specificity patient level (%) | False positives per case | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
With AI | Without AI | p | With AI | Without AI | p | With AI | Without AI | p | With AI | Without AI | p | |
Overall | 73.5 (68.3–78.2) | 70.1 (64.8–75.0) | 0.27 | 71.2 (65.3–76.6) | 66.3 (60.2–72.0) | 0.12 | 95.4 (93.8–96.7) | 93.3 (91.4–94.9) | 0.04 | 0.06 (0.05–0.08) | 0.09 (0.08–0.11) | 0.005 |
Physicians | 78.4 (71.3–84.5) | 73.5 (66.0–80.1) | 0.03 | 75.0 (66.7–82.1) | 66.7 (57.9–74.6) | 0.04 | 95.1 (92.6–96.9) | 93.2 (90.4–95.4) | 0.2 | 0.08 (0.06–0.11) | 0.11 (0.08–0.14) | 0.01 |
Neuroradiologist | 87.0 (75.1–94.6) | 87.0 (75.1–94.6) | 0.37 | 81.8 (67.3–91.8) | 72.7 (57.2–85.0) | 0.39 | 93.7 (88.3–97.1) | 89.4 (83.2–94.0) | 0.18 | 0.1 (0.06–0.15) | 0.1 (0.09–0.20) | 0.06 |
Radiologist | 79.6 (66.5–89.4) | 74.1 (60.3–85.0) | 0.61 | 79.5 (64.7–90.2) | 75.0 (59.7–86.8) | 0.62 | 96.5 (91.2–98.8) | 98.6 (95.0–99.8) | 0.37 | 0.03 (0.01–0.06) | 0.02 (0.01–0.05) | 1 |
Resident | 68.5 (54.4–80.5) | 63.0 (48.7–75.7) | 0.13 | 63.4 (47.8–77.6) | 52.3 (36.7–67.5) | 0.13 | 95.1 (90.1–97.8) | 91.2 (85.7–95.6) | 0.23 | 0.12 (0.08–0.17) | 0.17 (0.12–0.23) | 0.05 |
Student A | 70.4 (56.4–82.0) | 62.3 (48.7–75.7) | 0.39 | 70.5 (54.8–83.2) | 70.5 (54.8–83.2) | 1 | 96.8 (91.0–98.4) | 94.6 (88.3–97.1) | 0.58 | 0.04 (0.02–0.08) | 0.09 (0.05–0.14) | 0.45 |
Student B | 68.5 (54.4–80.5) | 59.3 (45.0–72.4) | 1 | 65.6 (50.1–79.5) | 68.2 (52.4–81.4) | 1 | 97.2 (92.3–99.2) | 92.3 (87.4–96.6) | 0.08 | 0.04 (0.02–0.08) | 0.08 (0.05–0.13) | 0.18 |
Student C | 66.7 (52.3–78.9) | 57.4 (43.2–70.1) | 0.36 | 65.9 (50.0–79.5) | 59.1 (43.2–73.4) | 0.61 | 94.4 (89.2–97.5) | 93.4 (88.3–97.1) | 1 | 0.06 (0.03–0.11) | 0.06 (0.03–0.11) | 1 |
Data in parentheses are 95% confidence intervals. AI artificial intelligence