Simulation · Interaction design
Grand Theft Health
I built this pandemic strategy game in R Shiny so that people could feel the trade-offs of running a country through an outbreak. You pick one policy a week while a disease model spreads infection across a live world map.

- Role
- Independent project: design and development
- Methods
- R, Shiny, Leaflet maps, Stochastic SEIR model, Scenario design
- Year
- 2025
The idea
Reading about epidemiology is very different from having to make the calls yourself, so in Grand Theft Health you’re in charge of a country during an outbreak and you have to live with the trade-offs. The policies that slow the disease aren’t always popular, and how popular you are decides how many people actually follow the rules.
How it plays
You pick a country and play week by week, choosing one policy card each turn and managing vaccines once they show up. Each week you only get a few cards to choose from, so what’s on offer shapes what you pick. Some cards are reasonable, like a mask mandate, and some are ridiculous on purpose, like replacing masks with inspirational quotes.
Every card changes how fast the disease spreads and how much the public likes you. You can watch what happens on a world map and on a dashboard that tracks cases, deaths, hospitalizations, hospital beds, GDP, unemployment, approval and vaccinations.
The model underneath
Every week, a disease model updates every country using real country data. It’s a stochastic SEIR model, which means it has some randomness built in and tracks people as susceptible, exposed, infected or recovered.
Infection spreads inside each country, across land borders and through air travel. Compliance depends on public approval, and more compliance means less spread. The number of doctors per 1,000 people and a country’s health spending change how deadly the disease is, and hospital overload adds extra deaths once the number of people needing hospital care outruns the beds.
The model took a long time to get working, because it kept producing impossible numbers, like more sick people than a country has people, until I added checks on every step. It’s a learning tool, so it shouldn’t be read as a forecast.