Skip to main content
. Author manuscript; available in PMC: 2025 Apr 19.
Published in final edited form as: Radiol Artif Intell. 2025 Mar;7(2):e230506. doi: 10.1148/ryai.230506

Table 5:

Model Classification of lung cancer risk by cancer subtype.

NLST-test-nodules (n malignant=147, n benign=749) Consortium-A (n malignant=238, n benign=126)
Model SCLC (n=18) NSCLC (n=119) SCLC (n=39) NSCLC (n=194)
Mayo 0.810 [0.808, 0.812] 0.774 [0.772, 0.776] 0.683 [0.681, 0.685]
Brock 0.792 [0.790, 0.794] 0.794 [0.792, 0.797] 0.688 [0.687, 0.690]
Liao et al. 0.683 [0.680, 0.687] 0.770 [0.768, 0.771] 0.617 [0.614, 0.620] 0.688 [0.686, 0.690]
Sybil 0.728 [0.723, 0.733] * 0.899 [0.897, 0.900] * 0.701 [0.698, 0.703] 0.701 [0.699, 0.702]
DLSTM 0.663 [0.658, 0.668] 0.808 [0.806, 0.809] 0.730 [0.726, 0.735] 0.754 [0.751, 0.757]
TDViT 0.707 [0.702, 0.711] 0.771 [0.769, 0.773] 0.667 [0.661, 0.673] 0.760 [0.757, 0.763]
DLS 0.659 [0.654, 0.664] 0.792 [0.791, 0.794]
DLI 0.904 [0.901, 0.907] * 0.853 [0.851, 0.855] *

Note.—Data are reported as bootstrapped mean area under the receiver operating characteristic curve [95% CI].

*

result was significantly different compared to each other method in the column for p<0.01

prohibitive class imbalance (n=5)

Missing demographic, smoking history, COPD, or CYFRA covariates