VisCareTrails: Visualizing Trails in the Electronic Health Record with Timed Word Trees, a Pancreas Cancer Use Case

As a mandate in the 2009 ARRS act, all US health care systems are moving toward electronic health record (EHR) systems to capture and store patient data. The EHR is a rich source of health information about individual patients and/or populations. The ability to analyze and identify meaningful patterns in this data has the potential to produce important knowledge. Yet, there is still a considerable gap between what answers are captured in this record and what answers can be effectively extracted from it. To reduce this gap, more intuitive ways of posing questions and obtaining answers are needed. In this paper we present VISCARETRAILS, a system based on timed word trees visualization that summarizes event paths relative to a given root event and are obtained through a simple drag-and-drop user interface. These summaries visually convey information about the nature, frequency and average timing of the event paths, and serve as a natural starting point to obtain further details and compare different paths. We apply VISCARETRAILS in a dataset of pancreatic cancer patients to illustrate its effectiveness.