Identifying Mentions of Life Stressors in Clinical Notes

The impact of social determinants on individual health has increasingly been recognized. In this study, we aimed to understand the representation of stressful life events occurring in clinical reports generated outside of mental health specialties. We present a conceptual schema for the representation of stressful events and associated linguistic attributes within free-text clinical notes. We also evaluated existing "off-the-shelf" clinical Natural Language Processing (NLP) systems for the detection of stress related concepts in clinical notes. We found that mentions of stressful events are prevalent even in non-mental health specialties, and that capturing details of stress mentions is challenging. Our results further indicate that existing NLP systems can serve as a reasonable starting point for developing models trained specifically for extracting stress associated information from clinical narratives.