The infectious disease forecasting discipline has grown exponentially over recent years, with numerous methods developed for application to both existing and emerging pathogens. Given the heterogeneous nature of recent epidemics and uncertainties in key epidemiological characteristics, such as observed with the unprecedented 2022- 2023 mpox epidemic and HIV pandemic, developing robust forecasting methodologies is essential to ensure timely and accurate forecasts. Additionally, continuous evaluation and refinement of applied methodologies are critical to advancing the discipline. Therefore, we present three studies applying and evaluating the versatile ensemble nsub- epidemic framework in the context of the 2022-2023 mpox epidemic on multiple geographic scales and the HIV epidemic in the United States.

In Study One, we employed publicly available HIV diagnosis data to produce multiple prospective forecasts through 2030 of incident HIV diagnoses overall, by region, race and ethnicity, and transmission type for the United States. We utilized the ensemble nsub- epidemic framework to conduct all forecasts. In Study Two, we conduct multiple short-term prospective and retrospective forecasts of incident mpox cases during the growth phase of the first wave of the epidemic globally and for Brazil, Canada, England, France, Germany, Spain, and the United States using the ensemble n-sub-epidemic framework. Additionally, we evaluated both the model fits for all forecasts and forecast performance of the retrospective forecasts via mean absolute error (MAE), mean squared error (MSE), 95% prediction interval coverage (PI), and weighted interval score (WIS). Finally, in Study Three, we evaluated and compared the retrospective forecast performance of two ensemble sub-epidemic frameworks, including the n-sub-epidemic framework, against four established statistical models for the initial wave of the mpox epidemic. We examined the performance of the models globally, and for Brazil, Canada, France, Germany, Spain, the United Kingdom, and the United States. We evaluated our results using the same performance metrics applied in Study Two.

While previous works validated the ensemble n-sub-epidemic framework in the context of respiratory infections (i.e., COVID-19), our findings extend its utility to pathogens spread primarily through intimate and sexual contact. The results of each study illustrate the framework's flexibility and precision, thus underscoring its value as a versatile real-time forecasting tool.