In this paper, we argue for setting forecasting accuracy as an explicit goal of epidemiological modeling. Specifically, we propose a public, ongoing evaluation program for measuring the forecasting accuracy of epidemiological models. The program should be designed by all stakeholders: PH decision makers, practitioners, funders and researchers. The evaluation should be based primarily on forecasting of events to occur in the coming 6-12 months, such as the dominant strain in the next epidemic wave, the number of documented animal-to-human transmissions, number of respirators used, etc. Based on the first author’s experience with designing standardized evaluations in other fields, we recommend the following process:
Convene a planning workshop to decide on focus and general guidelines for the evaluation.
Establish a governing body to oversee the evaluations on an ongoing basis.
Meet annually to analyze the results and revise the evaluation program.