Why Monte Carlo is Making You Work From Home
I am sure all of us have seen the curve shown below. This is the curve that is causing you to work from home right now. Most of us understand what this curve is about. The curve shows how many projected cases (projected being the key term) of Coronavirus infections would be recorded since the time of person infection in the general population. The Y-axis here is the number of cases recorded daily and the X-axis is time since the first infection. The pink curve here shows what would happen if we took absolutely no precautions and let the virus run its natural course. Meanwhile, the yellow curve shows what happens if we take all the precautions, for example, social distancing working from home all the fun stuff that we’re dealing with right now.
This powerful visual has been used by the CDC to convince all of us to stay away from places where people congregate. Somehow the CDC has come to the conclusion that social distancing will reduce the number of cases and overall health effects. How does the CDC come to this conclusion though? How are they so confident in their projections?
How Do They Know?
It is due to our good old friend Monte Carlo. If you want to get into the weeds of how they went about doing this, here are some articles from the CDC, Nature and a scholarly article from medRXiv that you will find interesting –
https://wwwnc.cdc.gov/eid/article/26/6/20-0233_article
https://www.nature.com/articles/s41421-020-0148-0
https://www.medrxiv.org/content/10.1101/2020.01.23.20018549v2.full.pdf
The nature article here is probably the most interesting read.
Here is my attempt at a simple explanation — The scientists at the CDC have been running simulations based on the models they have from the spread of previous infectious diseases (the SEIR model) and the data they have from how COVID-19 has spread. Armed with this information, the scientists varied what they call the basic reproduction number, R0. R0 is a representation of the number of new infections that would be generated by each infected person. They ran multiple Monte Carlo (to be precise Markov chain Monte Carlo) simulations with the various values for R0 to determine the estimates for what the spread of the disease would look like. The results are shown below in the figure taken from the Nature article.
They used Monte Carlo to not just figure out rates of spread, but also the Risk of Death. In fact most of the figures you hear about how many folks will be infected and could potentially die from this pandemic have been derived by running Monte Carlo simulations.
Math is Easy, Action is Hard
Ok, with the help of our models, inputs and these Monte Carlo simulations, we know that this is going to be a pandemic. Now what? We cannot control how ‘infectious’ the disease is. If folks encounter someone who has the disease, it is likely to get transmitted. Our Monte Carlo simulations are telling us that we need to do something otherwise our HealthCare systems are going to get overwhelmed. This is the tough part — Math is easy, action is hard. Luckily most of the world governments took immediate action and asked folks to practice social distancing. It is impressive how quickly folks have accepted these directives. There is an interesting story about social distancing and the Spanish flu from 1918, as it played out in St Louis vs Philadelphia — https://qz.com/1816060/a-chart-of-the-1918-spanish-flu-shows-why-social-distancing-works/.
The power of tools like Monte Carlo simulations is that they can help model scenarios and raise flags pretty early. Unfortunately, these tools are useless if we don’t take action based on the flags being raised. Monte Carlo Simulations can help us get an idea of things going off the rails while we can do something about it. It is up to us to do something about the concern these tools are raising and put some ‘distance’ between ourselves and failure. This is the tougher part of the equation. Doing the Monte Carlo and modeling math is easy, action is hard.
Yes, because of all this Monte Carlo is making you work from home.
Other applications
Almost every modern analysis that has uncertainty about the future, uses Monte Carlo to come up with projections. Here are the applications as listed by Wikipedia –
4.3 Climate change and radiative forcing
