willgreen93/attenuated-renewal

Forecasting emerging infectious disease outbreaks is particularly challenging in populations where transmission occurs on highly overdispersed contact networks, such as men who have sex with men (MSM). In these settings, early epidemic growth is often driven by a small number of highly connected individuals, leading to rapid early acceleration followed by structural attenuation as high-degree nodes are depleted. This repository implements and evaluates a suite of short-term epidemic forecasting models, with a focus on attenuated renewal processes that explicitly account for this early depletion. Using both large-scale stochastic simulations on synthetic MSM-like networks and retrospective forecasts of the 2022 mpox outbreak in multiple cities, the code demonstrates that sigmoid-attenuated renewal models provide more stable and reliable forecasts than standard renewal or purely statistical approaches.

Keywords