COVID-19 Modeling | One Year Later: Head of COVID-19 Modeling

IHME Director Chris Murray reflects on what we've learned since the first projections model.

Transcription: We've learnt an awful lot. We've learnt that you need to keep adapting the model as the science advances, and our understanding of COVID. We're now on our, essentially, fifth iteration of our model and we've progressed from, you know, learning about the transmission dynamics of COVID to having to add vaccination, when we were lucky enough to see vaccines coming into the model, to thinking about better ways to triangulate on infection rates as we get more and more people infected, and that becomes a bigger part of the discussion around models, to now in the last month, a real focus on building variants into the model, and dealing with this very thorny problem of reduced or very low levels of cross-variant immunity in some of the evidence that we're seeing, both from South Africa and from Amazonas in Brazil. 

So we've gone through those various iterations of the model. We have also learned about the critical ingredients that go into models, which is the data. You've got to have high quality information on cases, on deaths, on hospitalization, and it's been a long, constant journey, both getting the right data, persuading different governments, including the US government, to release the data, for example on hospitalization, and fixing all the problems that come with data, that are due to lack of testing or to lags in reporting over holiday periods, or snow storms in Texas, a constant set of challenges on the data front which you really have to grapple with if you want to get the models right.