Modelling African Swine Fever Transmission and Epidemiology: A Scoping Review of Mechanistic, Statistical, and Machine Learning Approaches

African swine fever (ASF) is a viral disease of domestic and wild pigs that has re-emerged as a major transboundary disease. Modelling using mechanistic, statistical, and machine learning (ML) approaches plays a key role in understanding ASF transmission and informing disease control, but the literature remains fragmented. To synthesise global ASF modelling efforts, we systematically reviewed studies applying these three approaches. We examined temporal and geographic trends, modelling objectives, explanatory variables, and model evaluation practices. A total of 151 papers published through 2024 met the inclusion criteria. Mechanistic (54.3%) and statistical (40.4%) approaches predominated, whereas ML (9.3%) was increasingly applied in recent years. Mechanistic models were primarily used to assess control strategies (48.8%) and transmission drivers (41.5%), statistical models to identify risk factors (63.9%) and spatiotemporal spread (32.8%), and ML for environmental suitability modelling (64.3%). Most were published from 2011 (99.3%) and focused on Europe (43.0%) and Asia (26.5%). Model evaluation remained inconsistent, with mechanistic papers frequently lacking model output uncertainty quantification (47.0%) and statistical papers often omitting model adequacy assessment (49.2%) and assumption checking (50.8%). Overall, ASF modelling approaches have developed complementary methodological roles, while geographic underrepresentation, limited representation of some transmission pathways, and inconsistent model evaluation remain important gaps.