Table 1 The comparison of causal variable determination for the PORTEC dataset among ICP, NICP and the proposed method
From: Automated causal inference in application to randomized controlled clinical trials
ICP | NICP | Proposed (warm-up) | Proposed (complete) | |
|---|---|---|---|---|
Causal determination of P | ||||
Pathological | ||||
Myometrial invasion | No | Yes | 55.33% | 52.13% |
Grade | No | Yes | 62.72% | 60.71% |
LVSI | No | Yes | 60.68% | 59.41% |
Sanity check | ||||
Tissue area | No | Yes | 49.31% | 35.96% |
Patient ID | No | Yes | 50.93% | 20.61% |
Causal determination of PM | ||||
Pathological | ||||
Myometrial invasion | No | Yes | 54.62% | 53.16% |
Grade | No | Yes | 60.44% | 60.42% |
LVSI | No | Yes | 62.13% | 60.87% |
Molecular | ||||
L1CAM | No | Yes | 60.27% | 60.43% |
POLEmut | No | Yes | 49.91% | 50.02% |
MMRd | No | Yes | 49.20% | 49.25% |
p53abn | No | Yes | 64.67% | 63.92% |
Sanity check | ||||
Tissue area | No | Yes | 49.77% | 24.94% |
Patient ID | No | Yes | 48.98% | 17.48% |
Causal determination of PMI | ||||
Pathological | ||||
Myometrial invasion | No | Yes | 55.22% | 53.84% |
Grade | No | Yes | 59.39% | 59.74% |
LVSI | No | Yes | 61.88% | 60.56% |
Molecular | ||||
L1CAM | No | Yes | 59.91% | 59.92% |
POLEmut | No | Yes | 50.08% | 50.46% |
MMRd | No | Yes | 49.10% | 49.40% |
p53abn | No | Yes | 63.82% | 62.91% |
Immune | ||||
CD8+ cell density | No | Yes | 56.33% | 56.14% |
Sanity check | ||||
Tissue area | No | Yes | 49.58% | 21.12% |
Patient ID | No | Yes | 49.18% | 22.89% |
