To The Top

Characterizing Physician EMR Workflow With Clustering and Hilbert Space Filling Curve Visualizations


Tim Green






2206A Student Center

Understanding sequences of events is paramount to developing a picture of how clinicians interact with other clinicians, patients, and Electronic Medical Records (EMR).  Common methods of assessing clinical workflow include qualitative methods such as video recording, direct observations, or directed user experience testing in a controlled environment. Studies such as these are designed to help understand traditional workflow concepts such as time and motion efficiency, interpersonal communication, and information needs.  Applications for EMR workflow include measuring productivity to help target training, and dynamic EMR experiences based on predicted actions.  Computational, or quantitative, methods can also be applied, such as sequential pattern mining, clustering, workflow mining, and graph-based mining.  Visualization techniques can also help in discovering patterns; however, the combination of longitudinal and multi-variate data can be challenging to visualize. Approaches to these challenges generally favor the use of motion charts to represent the passage of time while visualized elements represent other attributes and change over time.  In this talk, I will present a clustering methodology for grouping similar workflow patterns and visualizing workflow patterns efficiently through the use of space filling curves, such as the Hilbert Curve.