Inspiration
I had came to the hackathon originally wanting to base my project on public health but was temporarily discouraged by the amount of research it would require. After registration, when I began talking to sponsors - algo trading seemed to be a common theme and this caused me to look into Decentralised Finance and the possibility of doing a related project but I felt a bit lost as I didn't know much about the field and also it seemed really shady and not like something I actually wanted to look into.
After hearing about the challenges, I became inspired by the 'Best of Data for predictions and decision making" and had been previously investigating biosecurity and reading a few books on pandemic spread - this inspired me to make an interactive modelling platform in the hopes of serving as a tool to assess the potential for an outbreak of a disease to escalate into something more concerning.
What it does
There are a plethora of parameters available for you to configure to a disease of your choice - in practice, this would be a novel pathogen or a specific incidence of a known disease. After this is done, the model will provide a range of outputs that can be browsed and looked at. These include but are not limited to doubling time, daily growth, the estimated success of certain measures and
How I built it
Claude was used a lot - to be transparent my technical background is currently quite weak and Claude was a big help. I tried my best to still 'vet' the code - it was quite laborious copy and pasting it all into chatGPT and do my best to spot any potentially problematic errors. After Claude had provided the code, I uploaded it as a GitHub repo and used Streamline to launch it as an app.
Challenges I ran into
Certain model assumptions were a bit dubious and had to be revised - a big example was that Claude assumed the arbitrary value of 2x for the death multiplier of a hospital that became overwhelmed. Using a study published on English hospitals during COVID, I chose to replace the 2x with the multiplier that they calculated which was 1.13x, to make a more accurate model.
Accomplishments that I'm proud of
The logical reasoning behind the model is sound - a lot of improvements have been made since the first stage. My biggest thing is that the model actually makes sense, it is not yet capable of producing a timeline capable of emulating the natural progression of a typical outbreak but
What I learned
What's next for Outbreak Simulator
There are many
Log in or sign up for Devpost to join the conversation.