Simulate Anything

Simulate Anything Platform with a Simulation on the Impact of Social Distancing
Simulate Anything Platform with a Prediction of the Next 7 Day Case Count
Simulate Anything Platform Showing How R0 Week over Week Impacts Cases
Simulate Anything Platform Reproducing Expected Hospital Resource Shortage

Inspiration

My PhD advisor at Carnegie Mellon was among the first to think about generic simulations platforms and libraries that can be used across all biological modeling. Contrast this with specialized simulations such as those in fluid dynamics / airflow etc.

With COVID-19, there are a lot of people trying to build simulations and illustrate a point. However, it is impossible to reproduce the simulations or at least extremely hard to. One needs to go find the source code, rebuild it, re-run it and hope it produces the same results. Then if you wanted to change the simulation or expose new parameters, you have to fork the code and then host it yourself.

The wish that I have is that people can write simulations in a more standard form where we can collaborate, share and combine the source code, execution platform and results into one package. This can then easily be reviewed, modified and re-shared seamlessly.

What it does

simulateanything.com is a platform where people can write simulations very quickly and share them widely. We have kicked it off with a few simulations. The idea is to build a scratch pad for simulations quickly.

We have a simulation for predicting the impact of social distancing on the growth of the corona virus cases.
We have a prediction of the next 7 days case count data based on the recent history.
We have a simulation on how changing R0 on a week by week basis changes the future case count.
We reproduce the simulation of a published work on the stress on the healthcare system (beds, ICU, ventilators) and the expected shortage.

Again the point is not the individual simulations itself but that the platform significantly lowers the bar to write a simulation but also gives people power to run (or modify it) it themselves.

How I built it

We created a completely new platform where people can easily build and share simulations. The stack itself was written on node JS and all the execution is done on the client side (on the browser) and so we do not need significant simulation cycles on the server side (it is a hackathon :))

Challenges I ran into

As part of it the development technically the editor is a javascript IDE. However, I am not super happy with debugging capabilities of the editor at the moment. The line numbers on the console errors are lost in translation since the JS is inserted into existing client side code.

Accomplishments that I'm proud of

It feels like there is a need for a platform that can run simulations for anything. It would be so much nicer if a lot more simulations talked the same language and everyone can agree or even better disagree on the simulation based on reality than saying a simulation was bogus or they do not expose the right parameters etc.

What I learned

Dark mode is always better than non-dark mode :)

What's next for Simulate Anything

Write a couple more simulations. More importantly, share with others to ensure that this is something that would be useful in the long run. There are some famous simulations such as "Parables of Polygons" but these are always written from scratch over and over again. It would also be great to be able to undertake a significant simulation for a real enterprise COVID-19 use case than a toy proof of concept scratch pad of simulations.