[Return to Previous Page]
- Thesis Proposal Defense - Gregory Hruby
- Date : September 5th, 2013
Time : 3:00PM - 5:00 PM
Location : VC-5 Conference Room
Event Speaker : Greg Hruby
Understanding and Structuring the Clinical Data Query Mediation Process
Access to large amounts of clinical data made available by the electronic health record (EHR) is a costly barrier to computational reuse of EHR data for patient-centered outcomes research and comparative effectiveness research. To facilitate EHR data access, many institutions have integrated EHR data in a centralized data repository to better facilitate electronic data access for purposes other than clinical care. Additionally, dedicated data analysts or self-service query tools (QT) have been employed to provide access to these data for medical researchers. Often the QT tools are not sophisticated enough to support complex clinical data queries. For complex data request, a data analyst and medical researcher utilize a biomedical query mediation (BQM) process. Through an iterative consultation process, the medical researcher’s data request (often complex data elements and temporal or logical constraints attached to those elements) is transferred to the data analyst, who can translate such requests into an executable database query. BQM is arbitrary, as no formal process exists for data analysts to acquire the necessary skill sets. Additionally, BQM may involve inefficiencies and vague understandings of the information needs of the medical researcher. In order to facilitate the increased data requests brought on by researchers conducting comparative effectiveness research with electronic data, data analysts will need new tools to improve the efficiency of BQM.
The proposed research intends to understand, model, and structure BQM. I will accomplish these goals by developing and evaluating a guideline model to optimize BQM. I will vet the guideline model by measuring improvements in BQM efficiency, subjective measures from data analysts (perceived usefulness, usability, satisfaction), as well as objective measures such as number of iterations for query reformulation, task complexity and coverage, and data retrieval performance.