Figure 2. Receiver Operating Characteristics Curve Comparing Lung Cancer Prediction Models with Interaction Terms Among Combined (2a), Histoplasmosis-Endemic (2b), and Non-Endemic Cohorts (2c).

Figure 2a displays the apparent AUCs for lung cancer prediction models among the combined cohort (n=327). The combined biomarker model including Mayo, Histoplasmosis EIA, and radiomics exhibits the greatest AUC. Figure 2b displays apparent AUCs for lung cancer prediction models among the Histoplasmosis-endemic cohort (VUMC and UoC, n=157). The combined biomarker model including Mayo, Histoplasmosis EIA, and radiomics exhibits the greatest AUC. Figure 2c displays apparent AUCs for lung cancer prediction models among the non-endemic cohort (UPMC and DECAMP, n=170). The combined biomarker model including Mayo, Histoplasmosis EIA, and radiomics as well as the model including Mayo and radiomics exhibit the greatest AUCs.

AUC, Area under the receiver-operating characteristics curve

VUMC, Vanderbilt University Medical Center

UoC, University of Cincinnati

UPMC, University of Pittsburgh Medical Center

DECAMP, Detection of Early Lung Cancer Among Military Personnel