Using network graphs to visualize changing documentation styles in an oncology practice before and after opennotes implementation

摘要

OpenNotes is an international effort that provides patients with open access to their electronic health records (EHR). Whether or not OpenNotes has had any detectable influence on provider clinical note documentation practices (e.g., word choice, sentence structure, note length, etc.), especially in “high-stakes” clinical situations such as cancer diagnosis and treatment, remains largely unknown. To investigate this possibility, we present a visually driven method to portray word-level changes in clinical documentation via network graphs. Using EHR data collected from Beth Israel Deaconess Medical Center (BIDMC) before and after OpenNotes implementation, we analyzed hematology-oncology notes for changes in word co-occurrences. We have portrayed this data in a variety of network graphs in which words of interest are the nodes, and the edges connecting them depict the words' co-occurrences. By representing our data through network graphs, we have derived novel findings about the BIDMC hematology-oncology notes. Our preliminary approach towards OpenNotes portrays how network graphs may be a useful tool in unveiling word-level changes to clinical documentation.