The application of formal multilayer networks (MLNs)—networks which contain multiple types of relationships, data types, or other additional features of complexity and connectivity—has recently become popular in many fields. Areas of study ranging from social science to biology have utilized MLNs to explore, describe, and analyze interconnected complex systems. MLNs have demonstrated both flexibility and practicality in investigating high dimensional data due to the structural ability to integrate different types of related data into one mathematical model. An exemplar of such data are patient care activity logs, produced by clinician interaction with the electronic health record (EHR), which contains data on clinicians, patient encounters, and the care activities performed as part of treatment. This has inspired the development of an MLN model which analyzes patient care activity logs, with the aim of evaluating clinical processes by identifying when differences in clinical relationships are systematically predictive of patient outcomes. This thesis presents an applied MLN methodology to answer the following question: During which care activities are groups of clinicians with highly successful relationships most likely to impact patient health outcomes? Evidence is presented in three papers that supports the hypothesis that the applied MLN methodology accurately identifies both 1) highly successful clinical relationships among providers and 2) the areas of care most associated with those relationships and outcomes. The first paper explores the necessity and effects of risk adjusting patient outcomes to ensure accurate evaluation of clinical relationships. The second paper describes the MLN network in further detail and applies the proportion of categorized clinical relationships (as measured by risk adjusted patient outcomes) as edge-weights representing the connection between patient care activities performed on encounters, with the aim of identifying tasks where differences in relationships are linked to patient outcomes. This model structure is verified with simulated data and validated with treatment and outcomes data of intracranial hemorrhage (ICH) patients. In the final paper, the MLN model is further refined with the aim of increased clinical interpretability. Further evidence of method validity is presented from the examination of Computed tomography (CT) notes for the documentation of communication with other physicians. These methods present a new approach to leveraging EHR data by applying an MLN framework to investigate outcomes data with patient care activity logs. The evidence presented in this thesis supports the future utilization of these methods in targeted process improvement investigations and interventions.

Creator

• Chandler, Ariel

DOI

• https://doi.org/10.21985/n2-g4nk-1604

Subject

• Information science
• Health sciences
• Health care management

Language

• en

Alternate Identifier

• etdadmin_upload_713264
• http://dissertations.umi.com/northwestern:14964

Keyword

• Hospital Operations
• Intracranial Hemorrhage
• Patient Outcomes
• Teamwork
• Electronic Health Record Data
• Networks

Date created

• 2020-01-01

Resource type

• Dissertation

Rights statement

• In Copyright