Sepsis prediction models have had limited success with implementation in the inpatient setting as they often do not integrate into the electronic health record or do not provide real-time recommendations that improve physician workflow and decrease cognitive load. We explore the solution of a finite state machine (FSM) by extending an FSM used for sepsis prediction from the emergency department to the inpatient setting. We evaluate the prediction model’s discrimination and the complexities of inpatient sepsis management.

Learning Objective: 1. Learn challenges and possible solutions to the development and implementation of sepsis prediction models in the inpatient setting.
2. Understand the flexibility in modeling and interpretation that finite state machines offer for implementation of prediction models.

Authors:

Sameh Saleh (Presenter)
UT Southwestern Medical Center

Samuel McDonald, UT Southwestern Medical Center
Mujeeb Basit, UT Southwestern Medical Center

Presentation Materials:

• Abstract/Manuscript - 1