Duke University and the Duke Center for Health Informatics (DHCI) have been awarded an $800,000 grant titled “Novel Visualization of Large Health Related Data Sets.”
W. Ed Hammond, PhD, the Director of DCHI, is the PI. Additional team members include: Jeff Ferranti, MD, MS, Chief Medical Information Officer and Vice President for Clinical Informatics, Associate Director for Duke Center for Health Informatics, Assistant Professor of Pediatrics, and Assistant Professor in Community and Family Medicine; Carl F. Pieper, DPH, Assistant Professor of Biostatistics and Bioinformatics; Vivian West, PhD, MBA, RN, Associate Director of Operations at the DCHI; and David Borland, PhD, Senior Visualization Researcher at The Renaissance Computing Institute (RENCI), The University of North Carolina at Chapel Hill.
The amount of information in Electronic Health Record (EHR) systems is growing rapidly with the inclusion of disparate forms of data from a number of new sources, (i.e., genomics and imaging data). EHR systems will continue to grow as more health care data is digitized. As data in EHRs grow, there is a need to understand what information and knowledge these large data sets represent. Visualization offers an opportunity to explore and understand large data in ways never before possible. Disciplines such as computer science, engineering, and genetics have developed visualizations to improve presentation and understanding of data. The health care provider community has not yet taken advantage of these methods, nor has it significantly explored the use of new visualization techniques to accelerate the understanding and use of health-related data.
This project will explore interactive visualization of large sets of health data to provide better understanding of what is in the data. Retrospective data queries from DEDUCE will be used to evaluate what information clinicians seek from health care data, identify what data elements and mixtures of data classes (e.g., laboratory data, demographic data, problems, therapies, physical examination data, or imaging data) are used in queries and what methods are used to analyze the results of those queries.
A matrix will be created of data visualization methods used with specific data elements from multiple classes and test visualization of mixed data classes (i.e., combining laboratory and imaging data in the same visualization). Various visualization techniques will be used to present the results of hypothesis-driven queries, and visualization methods will be developed and analyzed in which informational content from large databases will present or “discover” itself without specific hypotheses. Compressing petabytes of health care data representing many data elements into various groups of related data presented visually with an interface that allows the user to interactively explore the data elements is a novel approach. This visualization technique has the potential to detect causal relationships between various sets of data, which may lead to improved health care and save millions of dollars in health care costs.