Visualizing Patient Timelines in the Intensive Care Unit

Electronic Health Records (EHRs) contain a large volume of heterogeneous patient data, which are useful at the point of care and for retrospective research. These data are typically stored in relational databases. Gaining an integrated view of these data for a single patient typically requires compl...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2018-06
Main Authors: Levy-Lambert, Dina, Gong, Jen J, Naumann, Tristan, Pollard, Tom J, Guttag, John V
Format: Article
Language:English
Subjects:
Electronic health records
Laboratory tests
Mathematical models
Query languages
Relational data bases
Software reviews
Visualization
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Electronic Health Records (EHRs) contain a large volume of heterogeneous patient data, which are useful at the point of care and for retrospective research. These data are typically stored in relational databases. Gaining an integrated view of these data for a single patient typically requires complex SQL queries joining multiple tables. In this work, we present a visualization tool that integrates heterogeneous health care data (e.g., clinical notes, laboratory test values, vital signs) into a single timeline. We train risk models offline and dynamically generate and present their predictions alongside patient data. Our visualization is designed to enable users to understand the heterogeneous temporal data quickly and comprehensively, and to place the output of analytic models in the context of the underlying data.
ISSN:2331-8422