Retrospective evaluation of short-term forecast performance of ensemble sub-epidemic frameworks and other time-series models: The 2022-2023 mpox outbreak across multiple geographical scales, July 14 th , 2022, through February 26th, 2023.

In the face of many unknowns (i.e., transmission, symptomology) posed by the unprecedented 2022-2023 mpox epidemic, near real-time short-term forecasts of the epidemic's trajectory were essential in intervention implementation and guiding policy. As case levels continue to dissipate, evaluating the modeling strategies used in producing real-time forecasts is critical to refine and grow the field of epidemiological forecasting. Here, we systematically evaluate the performance of an ensemble -sub-epidemic and related sub-epidemic wave (spatial-wave) modeling frameworks against ARIMA, GAM, Prophet, and SLR models in producing sequential retrospective weekly (1-4 week) forecasts of mpox cases for the highest burdened countries (i.e., Brazil, Canada, France, Germany, Spain, the United Kingdom, and the United States) and on a global scale. Overall, the -sub-epidemic framework outperformed all other models most frequently, followed closely in success by the spatial-wave framework, GAM, and ARIMA models regarding average MSE, MAE, and WIS metrics. The -sub-epidemic unweighted model and spatial-wave framework performed best overall based on average 95% PI coverage, and we noted widespread success for both frameworks in average Winkler scores. The considerable success seen with both frameworks highlights the continued utility of sub-epidemic methodologies in producing short-term forecasts and their potential application to other epidemiologically different diseases.