Objectives: The goals of this study were to examine relationships among health literacy and outcomes for sub-populations identified within a large, multi-dimensional Omaha System dataset. Specific aims were to extract sub-populations from the data using Latent Class Analysis (LCA); and quantify the change in knowledge score from pre- to post-intervention for common sub-populations. Design: Data-driven retrospective study using statistical modeling methods. Sample: A set of admission and discharge cases, captured in the Omaha System, representing 65,468 cases from various health care providers. Measures: Demographic information and the Omaha System terms including problems, signs/symptoms, and interventions were used as the features describing cases used for this study. Development of a mapping of demographics across health care systems enabled the integration of data from these different systems. Results: Knowledge scores increased for all five sub-populations identified by latent class analysis. Effect sizes of interventions related to health literacy outcomes varied from low to high, with the greatest effect size in populations of young at-risk adults. The most significant knowledge gains were seen for problems including Pregnancy, Postpartum, Family planning, Mental health, and Substance use. Conclusions: This is the first study to demonstrate positive relationships between interventions and health literacy outcomes for a very large sample. A deeper analysis of the results, focusing on specific problems and relevant interventions and their impact on health literacy is required to guide resource allocation in community-based care. As such, future work will focus on determining correlations between interventions for specific problems and knowledge change post-intervention.
Objectives: The goals of this study were to examine the feasibility of using ontology-based text mining with CaringBridge social media journal entries in order to understand journal content from a whole-person perspective. Specific aims were to describe Omaha System problem concept frequencies in the journal entries over a four-step process overall, and relative to Omaha System Domains; and to examine the four step method including the use of standardized terms and related words. Design: Ontology-based retrospective observational feasibility study using text mining methods. Sample: A corpus of social media text consisting of 13,757,900 CaringBridge journal entries from June 2006 to June 2016. Measures: The Omaha System terms, including problems and signs/symptoms, were used as the foundational lexicon for this study. Development of an extended lexicon with related words for each problem concept expanded the semantics-powered data analytics approach to reflect consumer word choices. Results: All Omaha System problem concepts were identified in the journal entries, with consistent representation across domains. The approach was most successful when common words were used to represent clinical terms. Preliminary validation of journal examples showed appropriate representation of the problem concepts. Conclusions: This is the first study to evaluate the feasibility of using an interface terminology and ontology (the Omaha System) as a text mining information model. Further research is needed to systematically validate these findings, refine the process as needed to advance the study of CaringBridge content, and extend the use of this method to other consumer-generated journal entries and terminologies.
