Skip to main content
. 2024 Aug 13;15:6931. doi: 10.1038/s41467-024-51202-2

Fig. 3. Results for Use Case II—Pai et al.

Fig. 3

We replicated the findings of Pai et al. on the non-small cell lung cancer (NSCLC) Radiomics dataset and the NSCLC-Radiogenomics dataset. The linear model built on top of Pai et al.’s foundation model achieved an AUC of 0.64 on the NSCLC-Radiomics collection (N = 420) and of 0.65 on the NSCLC-Radiogenomics collection (N = 133) and significantly stratifying patients in the Kaplan–Meier analysis (P < 0.001 and P = 0.006 on the two cohorts, respectively). The high-risk group had a hazard ratio (HR) of 1.43 (95% CI 1.16–1.76, P < 0.005) compared to the low-risk (baseline) group on the NSCLC-Radiomics dataset, while on the NSCLC-Radiogenomics dataset, the HR for the high-risk group was 2.4 (95% CI 1.29–4.46, P = 0.01), consistent with the results reported in the original publication. Credit (icons): FlatIcon.