This is the weekly note for my project in Trinity Term 2025 at the University of Oxford, supervised by Professor David Gavaghan. The topic is modelling epidemic dynamics using Agent-based model (Epiabm) and SEIRD model in aid of policy making and disease control.

The task in week one is to learn, code up and fully understand the SEIRD model. See 1_SEIRD.

In week two, we look at the inverse problem of finding the posterior distribution of parameters, by fitting noisy data to the model and using PINTS. See 2_Inference.

In the following weeks, we try to reproduce the results of this paper: use the synthetic data generated by Epiabm to assess the performance of SEIR model under spatially heterogeneous setting. The workflow consists of three steps:

1. EpiGeoPop: generate standardized population configuration file for a certain region
2. Epiabm: generate epidemic simulation data based on the population configuration file
3. SEIRMO: model inference, generate SEIR estimated $R_t$ and other relevant plots

You can see full detail in 3_Geo_Hetero.