Toward a Generalized Model of Biomedical Query Mediation to Improve Electronic Health Record Data Retrieval

The electronic health record (EHR) is an invaluable resource for medical knowledge discovery. EHR data interrogation requires significant medical and technical knowledge. To access EHR data, medical researchers often rely on query analysts to translate their EHR information needs into EHR database queries. The conversation between the medical researcher and the query analyst is an information needs negotiation; I have named this process biomedical query mediation (BQM). There exists no BQM standard to guide medical researchers and query analysts to effectively bridge the communication gap between these medical and technical experts. The current practice of BQM likely varies among query analysts. This variation may contribute to the delivery of EHR data sets with varying degrees of accuracy. For example, a query analyst may return an EHR dataset that misrepresents the medical researcher’s information need or another query analyst may return a different EHR dataset to the medical researcher for the same information need. The process used to formulate the medical researcher’s information need and translate that need into an executable EHR database query may have severe downstream consequences affecting the reliability and quality of EHR datasets for medical research. This dissertation contributes early understandings of the BQM process and thereby improves the transparency and highlights the complexity of BQM by completing five studies: 1) survey the literature from other information intensive scientific disciplines to identify knowledge and methods potentially useful for BQM, 2) perform a review of existing tools and forms for assisting researchers in BQM, 3) perform a content analysis of the BQM process, 4) conduct a cognitive task analysis to detail a generalized workflow, and 5) develop an enriched concept schema to capture comprehensive EHR data needs. This dissertation employs extensive qualitative methods using grounded theory, expert interviews, and cognitive task analysis to produce a deep understanding of BQM. Additionally, I contribute a promising concept class schema to represent medical researchers’ EHR data needs to help standardize the BQM process.