Pathak, Winner2025-09-152025-09-152025-08-142025-08-15http://hdl.handle.net/1993/39367Infectious diseases spread rapidly and pose significant global health risks. Understanding the mechanisms of infectious disease transmission is essential for informing effective public health policy. To this end, individual-level models (ILMs) provide a flexible framework to incorporate covariate information when modelling. ILMs have been recently expanded to geographically dependent ILMs (GD-ILMs) to address spatially varying risk factors. Parameter estimation for these models is typically done using a Bayesian framework, which is computationally expensive. In this work, we propose to use a tree-based epidemic model classification method to bypass computationally intensive likelihood calculations. Specifically, we use a deep forest classifier and evaluate the predictive ability of the corresponding GD-ILMs to capture disease transmission dynamics. We validate this approach with simulated epidemic data and the UK 2001 foot-and-mouth disease outbreak, and compare its performance with that of the standard random forest classifier. Our findings suggest that the deep forest classifier generally outperforms the standard random forest classifier, as it uses an additional layer to extract more information from predictor variables.engInfectious DiseaseEpidemic ModellingEnsemble Learning