MPhys project: Modelling the role of anti-vaccination sentiment in disease outbreaks on a heterogeneous network
Hello! Thank you for taking an interest in my MPhys project. I created this code during the 2021/2022 academic year, and used its outputs to write my final Masters dissertation. That said, this repository is absolutely not in its final state - there are many code snippets to tidy, graphs to add, and more. Since my graduation, this has become a side project for me, so please don't expect progress to be swift!
This simulation was designed to model the interaction between two key facets of disease propagation: transmissability and vaccination uptake. While the simulation is designed to be non-specific to any particular disease, it currently uses some hard-coded parameters (R0 number, post-infection immunity time, etc) which were the best-known approximation of those values for COVID-19 as of March 2022. One application of this project is that this code can be altered for application to other infectious repiratory diseases, though I currently have no intention of adding alternative parameters myself.
During my Masters project investigations, the simulation produced the following main conclusions:
- Higher anti-vaccination sentiment in the population leads to an exponential increase in the number of disease cases observed during an endemic outbreak
- Higher anti-vaccination sentiment can also increase the likelihood that a disease will become endemic by up to 11%
- Those who do not choose to become vaccinated carry a higher burden of cases than their regularly vaccinated counterparts, the magnitude of which is directly correlated to their number of physical contacts.
There are three main systems which interact to affect disease outcomes in the model:
- A disease transmission model: this is a relatively simple SIRS model, where individuals are always in one of three states: susceptible to disease, infected with disease, or recovered from disease (a temporary immunity state which eventually reverts back to susceptible).
- A vaccination system: individuals within the simulation are regularly offered vaccination, which they may choose to take up or turn down. If vaccination is accepted, an individual will enter an immune state for an extended period (functionally identical to the recovered state in the SIRS model).
- An opinion system: each individual holds an opinion (either positive or negative) on vaccination, and can influence their social contacts to adopt their same opinion. If the opinion is negative at the time a vaccination is offered to an individual, they will decline the vaccine. Otherwise, they will become vaccinated.
Each individual in the simulation has a list of social contacts (to whom they can spread their opinion) and a list of physical contacts (to whom they can spread disease). These lists will contain overlap, as is likely in the real world. The data for these contact numbers was taken from the POLYMOD dataset (please see the Supplementary analysis folder for more information).
The full model code (used for analysis in my MPhys report and presentation) is accessed by running main.py. A pared-down, terminal user-friendly version of the program (the "public summary" component of my Masters Project) is available by running covid_game.py.
These scripts will import modules vaccination.py, network.py and voter_model.py from the modules folder, which contain auxillary functions which are used in the main simulation.
The simulation is designed for use from a console or terminal.
The user can see the progression of the simulation in real time, with two output options:
- Realtime statistics provided in a table
- Transmission/recovery events in a list
There is also the option to turn off terminal output.
The following parts of this repository still require work/updating to reach their final state:
- General cleaning of existing code and more in-depth commenting throughout