Epidemic Through Time

Screenshot from demo, Starting location set to London

Inspiration

Advancements in education and medicinal fields. Future potential for predictive models of virus outbreaks. Interest in large language models.

What it does

The project accepts parameters for a virus at specific times in history. You may place the virus anywhere in the world as long as it existed at the time. The virus will spread completely based off GPT's decisions. For example, it could be across the globe due to the inventions of airplanes in the modern era. However in the bronze age, it may be blocked by mountains.

How we built it

We split up into different groups. Someone handled functions that include equations used in real life to track the spread of viruses such as COVID-19. We prompt engineered GPT to produce the correct outputs and linked the agents. Finally the Unity frontend was handled by someone else to create an interactable environment.

Challenges we ran into

Interfacing with OpenAI's API and obtaining correct output formatting. Since we were making a virus that spread out throughout the world, the calls to chatgpt grew exponentially leading to some problems with procuring the correct outputs.

Accomplishments that we're proud of

We are proud of consistently work on something for a prolonged duration of time. Collaboration with motivated peers.

What we learned

Learnt how to communicate closely between team members working on different parts of the project. We learnt data collection through LLMs. We learnt how to utilise AI agents.

What's next for Epidemic Through Time

Hopefully we would be able to provide some insights on why certain decisions were made. For example, we could visually show the transport between countries. E.g: like how a virus could spread to America from Europe via a plane.