Models built to augment the physicians’ ability in empiric antibiotic prescription need to be evaluated on two factors: the accuracy of the model in covering the infection and if it suggests the narrowest-spectrum antibiotic that could cover the infection. Prescribing broad-spectrum antibiotics can have long term negative impacts such as increased resistance in the community. In this work, we explore the trade-offs of different weighting schemes to recommend an antibiotic from the output of different regression models.

Learning Objective: - Learn about building models to aid physicians in empiric antibiotic prescription
- Identify different criteria for evaluating the effectiveness of models
- Understand trade-offs and complexities when selecting between outputs of multiple models

Authors:

Protiva Rahman (Presenter)
The Ohio State University

Erinn Hade, The Ohio State University
Courtney Dewart, The Ohio State University
Yuan Gao, The Ohio State University
Courtney Hebert, The Ohio State University

Presentation Materials:

• Abstract/Manuscript - 1