Modeling COVID-19 from Imperial College London

The below is taken from excerpts of a group thread of entrepreneur marketing nerds in London.
It’s a private group, hence no names or contacts attributed to excerpts, but valuable reading given the explicitly-stated credentials of the bottom comment:

 


Re modelling, most of the scenarios from Imperial are pretty grim. All options massively overwhelm NHS capacity (red horizontal line).

image.png

 


 


 

Re the Imperial paper: Really well written imo and anecdotally what convinced Boris to change course and has also been taken to white house and will inform strategy there. For nerds, it’s worth reading the methodology section also. In a long-forgotten former life my PhD was on stochastic modelling of epidemics – and their approach is pretty much as good as it gets there: stochastic model, explicit spatial structure through demographics down to schools, parameters informed through what we know so far (such as incubation period, recovery rates, …) sensibly adjusted for UK/US realities, fitted to match the initial actual data we know, lots of sensible sensitivity analysis…

I went through the following takeaways while reading
  • (First graph above) Oh my god we’re so screwed, can I still get on a plane to Germany
  • (Second graph) Oh great, there’s a strategy that means the NHS won’t totally blow up
  • The implications of this will not blow over soon and may stay with us for a while (rebound after loosening restrictions, further waves, …), until there’s a vaccine (~18m)
I also like the (out there) hypothesis of this thing not being stable/effective outside a narrow temperature range