Fig. 1

Population selection diagram of the validation study. To validate the index classifier against the TNM 7th edition, a retrospective study was conducted with electronic health records (EHRs) of NSCLC patients in a tertiary referral hospital in Seoul, Korea. We identified a consecutive series of all patients who were newly diagnosed with NSCLC and staged between 2011 and 2015, covered under the national health insurance, and received cancer treatment in the hospital with their last visit more than 180 days from the end of observation. We excluded cases initially diagnosed at an external hospital, transferred to an external hospital, missing stage information, erroneous stage identified during the manual chart review process, received no treatment, missing age information, age of less than 20 years or more than 75 years, preexisting non-pulmonary cancer, or cardiopulmonary comorbidity (eMethods 2)