Preliminary Results of COVID Modeling

I want to take a moment to give an update on our COVID modeling effort. We have continued working on our machine learning model, and have been able to tune it to yield very good predictions. Instead of predicting case counts, we shifted to predicting R0 (or the rate of transmission).

Essentially, we tried to evaluate how different scenarios would result in different rates of infection. Here is a quick figure of how this works. On the y-axis the transmission rate (R0) which we are trying to predict. A R0 below 1 is desirable. The red line is empirical time series of what actually happened. The model is able to predict what actually happened very closely (within 0.1 most of the time). The other colors represent how different scenarios might change the rate of infection based upon other data. So for instance, quantifying how much relaxing social distancing will affect the infection rate. While we are still working to define each scenario better for the countries we are considering, the model is fairly accurate.

We are now working with some disease modelers to get this work published and hopefully can be used to improve other models and make decisions. If you are patient and interested in seeing how different scenarios might work, we have the code and model uploaded to this website. Once you click run, you will need to be a bit patient while the model computes, but it can show you how things might change. Once we refine it more, I’ll post another update.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.