We present a strategy for modeling proposed algorithms on recent patient data to optimize the development of a real-time pediatric sepsis detection tool. These tools test “fast-track” pathways whereby patients who already have sufficient charted information to trigger a sepsis huddle may skip prompted nurse assessments. By modeling the behavior of the tool on real historical patient encounters we aim to identify potential pitfalls before the resource-intensive phases of building and silent monitoring.

Describe the new knowledge and additional skills the participant will gain after attending your presentation.: The attendee should be able to assess the role of retrospective data analysis in the formulation of new clinical tools, and describe the strengths and limitations of retrospective analysis.


Amelia Drace (Presenter)
Oregon Health & Science University

Benjamin Orwoll, Oregon Health & Science University
Matthew Storer, Oregon Health & Science University
David Dorr, Oregon Health & Science University

Presentation Materials: