B1872
Title: Prediction models, robustness, and decision-making
Authors: Tyler McCormick - University of Washington (United States) [presenting]
Abstract: As methodology advances and software to implement complicated prediction models gets easier to use, a rise in settings is seen where some or all decisions are based on prediction models. A series of case studies are presented from global health, infectious disease epidemiology, and policy where prediction models appear to be an appealing stand-in for time-consuming, expensive (and also imperfect) data collection. In each setting, the potential utility of prediction models is evaluated, while also evaluating possible downsides of prediction errors and how various existing notions of robustness can (or cannot) offer solutions.
