The Biostatistics and Computational Biology Shared Resource (BCB-SR) at the UF Health Cancer Institute can now assist researchers with using Epic COSMOS, a large-scale, de-identified electronic health record database containing data from hundreds of millions of patients across contributing healthcare organizations worldwide, including UF Health.
Ada Wang, M.S., a biostatistician in the BCB-SR, recently acquired the COSMOS Data Model certification.
While general COSMOS users can use SlicerDicer to generate descriptive summaries from a limited COSMOS dataset, Cosmos Data Model-certified users are able to work with the full de-identified, individual-level COSMOS dataset through the COSMOS portal. This capability enables more advanced cohort construction and statistical analyses tailored to specific research questions.
Epic COSMOS includes a broad range of clinical and patient-level information, such as diagnoses, medications, laboratory results, encounters, vital signs, birth records, patient-generated health data and social drivers of health, including transportation and financial security assessments. With its large and diverse patient population, COSMOS provides a valuable resource for generating real-world evidence across a wide range of clinical research questions, including cancer research and studies of rare conditions.
SlicerDicer: COSMOS users can use SlicerDicer, Epic’s self-service cohort exploration tool, to explore COSMOS data and generate aggregate-level counts and descriptive summaries. It is useful for initial cohort exploration and feasibility assessment but provides more limited analytical capabilities than working directly with individual-level COSMOS data.
DSVM (Data Science Virtual Machine): COSMOS Data Model-certified users can work with de-identified, individual-level COSMOS data in the DSVM environment. This enables analysts to construct study-specific cohorts and datasets and perform more advanced statistical analyses using tools such as SQL and statistical programming software.
How BCB-SR Can Support Your COSMOS Research
Workflow:
- Research Question
- Feasibility Analysis
- Cohort Construction
- Validation
- Analysis
- Publication
UF Health Cancer Institute investigators interested in using COSMOS can collaborate with the BCB-SR throughout the research process. The team can help evaluate the feasibility of a proposed research question by reviewing available variables in the COSMOS data dictionary and exploring the data available for analysis. For feasible projects, BCB-SR can construct study-specific data marts by applying knowledge of the COSMOS Data Model, COSMOS infrastructure and SQL programming to define study cohorts and prepare analysis-ready datasets.
BCB-SR can also provide statistical expertise for the analysis and interpretation of COSMOS data, including descriptive analyses, data visualization, regression modeling, predictive modeling and other statistical methods appropriate to the research question. Investigators can work collaboratively with the BCB-SR on data validation, interpretation of findings and development of research products and publications.
