Development of real-time forecast modelling for lumpy skin disease outbreaks

Abstract Lumpy skin disease (LSD) is an emerging transboundary and vector-borne disease of cattle that has spread rapidly across East Asia and Europe. Generating reliable forecasts during the early stages of an outbreak remains challenging because of limited epidemiological data, uncertainty in vector transmission, and differences in livestock production systems across countries. We developed a real-time forecasting model for LSD that incorporates intra- and inter-farm transmission, vector dynamics, and Approximate Bayesian Computation Sequential Monte Carlo for iterative model fitting. We implemented and compared two distinct approaches for between-farm transmission involving force-based infectious pressure and stochastic infection probability, evaluating predictive performance with continuous ranked probability scores. We evaluated the model using both a simulated outbreak in Northern Australia and the 2024 outbreak in Japan. The model accurately captured spatiotemporal dynamics even when fit to only the first three weeks of outbreak data in both outbreak contexts. The infection probability approach had higher spatiotemporal accuracy than the infectious pressure approach. This study demonstrated how this model can accurately capture LSD outbreak dynamics while flexibly accommodating structural differences in national livestock husbandry systems. The framework provides information and insights for rapid outbreak response and serves to guide strategic decision-making across diverse national contexts.