Abstract
Social determinants of health (SDoH) are important factors associated with cancer risk and treatment outcomes. There is an increasing interest in exploring SDoH captured in electronic health records (EHRs) to assess cancer risk and outcomes; however, most SDoH are only captured in free-text clinical narratives such as physicians' notes that are not readily accessible. In this study, we applied a natural language processing (NLP) system to identify 15 categories of SDoH from a total of 10,855 lung cancer patients at the University of Florida Health. We aggregated the SDoH concepts into patient-level and assessed how each of the 15 categories of SDoH were documented in cancer patient's notes. To the best of our knowledge, this is one of the first studies to examine the documentation of SDoH in clinical narratives from a real-world lung cancer patient cohort. This study could guide future studies to better utilize SDoH information documented in clinical narratives.