Bridging Conversation Islands to Connect Healthcare: Introducing Unique Co-occurring Word Networks to Find Distinct Themes

Abstract

A common step in any modeling, study, or design project is gathering, reviewing, and making sense of relevant information. Common themes and variation are unearthed from text documents and shaped into usable forms to support the information needs of the project team. It can be a challenge to gather needed information from large numbers of texts and documents necessary to inform design decisions. Computational approaches can be useful in filtering and interpreting information. Unsupervised learning approaches like topic modeling can help to group similar texts and visualize potential themes. Just like grouping texts into similar themes, identifying distinctions can be helpful. This paper describes an approach for finding distinctions across texts and examples of using this information to inform decisions. Finding distinct themes can help to identify gaps and opportunities while guiding decisions and next steps. Identifying and bridging conversation islands across information sources and professions is another step towards managing complexity and connecting the dots of healthcare.