Additional file 1 of Automated detection of lung nodules and coronary artery calcium using artificial intelligence on low-dose CT scans for lung cancer screening: accuracy and prognostic value

Additional file 1: Table S1. Demographics of patients with and without lung nodules stratified by the AI and expert as well as expert CAC scores. Table S2. Comparison of risk factors and clinical attributes between patients with expert determined nodules, comparison of risk factors and clinical attributes between patients with AI determined nodules, and comparison of risk factors and clinical attributes between patients with CAC > 0 and CAC = 0. Table S3. Demographics and risk factors associated with pulmonary outcomes. Table S4. Demographics and risk factors associated with cardiac outcomes. Table S5. Simple logistic regression for parallel analysis of AI-volume and expert-volume for prediction of cardiac outcomes. Table S6. AUC and McFadden R2 for outcomes with and without AI components included in the model. Table S7. Summary statistics of Patients with False Positive Nodules. Figure S1. ROC curves for comparison of CAC AI-Volume and Expert-Volume for prediction of MACE. Expert and AI-Volume both excellently predict MACE. Figure S2. ROC Curves for comparison of CAC AI-Volume and Expert Volume for prediction of ACS/MI hospitalization in our study timeframe. Figure S3. ROC Curves for comparison of CAC AI-Volume and Expert Volume for prediction of percutaneous coronary intervention (coronary catheterization or stent placement) or coronary artery bypass graft operation. Figure S4. Root cause analysis of false-positive nodules. A. Logistic regression of having one false positive nodule as predicted by age. B. Logistic regression probability curve of false positive nodules as a function of age. C. True anatomic identities and relative frequencies of false positive nodule etiologies.