Table 1 Performance metrics of LunaCAM models in plasma samples
| Â | LunaCAM-S model | LunaCAM-D model | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Â | Â | Total | Negative | Positive | Sensitivity (95% CI) | Specificity (95% CI) | Total | Negative | Positive | Sensitivity (95% CI) | Specificity (95% CI) |
Plasma (model building set) | |||||||||||
|  | Cancer | 190 (training) | 16 | 174 | 91.6%(90.1–93.4%) | - | 190 (training) | 83 | 107 | 56.3%(52.6–59.9%) | - |
|  | Benign | 135 (test) | 57 | 78 | - | 42.2%(38.0-46.3%) | 135 (training) | 132 | 3 | - | 97.8%(97.2–99.1%) |
|  | Health | 188 (training) | 113 | 75 | - | 60.1%(56.7–63.3%) | 188 (test) | 166 | 22 | - | 88.3%(86.0-90.7%) |
Plasma (Independent validation set) | |||||||||||
|  | Cancer | 64 | 6 | 58 | 90.6%(88.2–94.1%) | - | 64 | 29 | 33 | 51.6%(45.1–56.9%) | - |
|  | Benign | 45 | 17 | 28 | - | 37.8%(30.6–44.4%) | 45 | 42 | 3 | - | 93.3%(91.7–97.2%) |
|  | Health | 63 | 39 | 24 | - | 61.9%(56.0–68.0%) | 63 | 53 | 10 | - | 84.1%(80.0–88.0%) |
Plasma (Total) | |||||||||||
| Â | Cancer | 254 | Â | Â | Â | Â | 254 | Â | Â | Â | Â |
|  | I | 108 | 14 | 94 | 87.0%(83.7–90.7%) | - | 108 | 55 | 53 | 49.1%(44.2–53.5%) | - |
|  | II | 38 | 2 | 36 | 94.7%(93.3–100.0%) | - | 38 | 15 | 23 | 60.5%(53.3–70.0%) | - |
|  | III/IV | 108 | 6 | 102 | 94.4%(93.0-96.5%) | - | 108 | 44 | 64 | 59.3%(54.7–64.0%) | - |
|  | Health | 251 | 152 | 99 | - | 60.6%(57.5–63.5%) | 251 | 219 | 32 | - | 87.3%(85.5–89.5%) |
|  | Benign | 180 | 74 | 106 | - | 41.1%(37.5–44.4%) | 180 | 174 | 6 | - | 96.7%(95.8–97.9%) |