University of Melbourne has released an app designed to model how fast COVID-19 might be spreading in various countries, including Australia.

See the application here.

Code is available on the Github.

Associate Professor Ben Phillips explains the model in a blog post:

It is difficult to estimate the total number of actual cases of COVID-19, but it is important to try to do so. We need to appreciate the potential scale of the pandemic if, as a global community, we are to tackle it effectively.

Accordingly, we have made a rough estimate of the proportion of cases that are being detected in each country so that we can work backwards and generate a figure for total cases.

The method assumes that there is widespread community transmission, that deaths won’t go unnoticed, that the case fatality rate of symptomatic people is about 3.3 per cent and it takes about 17 days for people who are going to die from the virus to pass away.

These numbers come from published analyses of the Wuhan outbreak, and where there is uncertainty we have chosen values that give us a larger detection probability.

Under these assumptions, we can then look at the number of deaths occurring in a five-day period, and estimate the number of infections required to generate these deaths based on a 3.3 per cent fatality rate.

Finally, we can compare that to the number of new cases actually detected in the five-day period 17 days earlier to give us an estimate of the proportion of actual cases that were detected 17 days ago.

This can then give us an estimate of the total number of cases, confirmed and unconfirmed.