Covid-19 – a layman’s model

The models are being talked about by everyone, all the time. The one by Imperial college, UK researcher Neil Ferguson has heavily influenced the public policy in US and dominated the media.

So what features these models rely on? How does their predictions have been so up and down? How does mitigation efforts impact the predictions by these models? What does Dr. Fauci means when he says a model is heavily dependent on the assumptions inherent in it?

Kaggle as expected has some data out there for analysis and model building. However, I try to look at the factors that could be building blocks of such models, based on common sense understanding of the virus spread.

It is generally assumed there is a patient zero, meaning only one human gets the infection from bats as it is now well established that the virus originated in bats. Now, once infected, a person may have following states

  • Stay asymptotic and recover , not knowing ever they were infected
  • Stay asymptotic for a short time and then show mild symptoms, eventually recovering without serious medical intervention and a test
  • Not withstanding initial symptoms, person has serious issues and ends up having a test and medical intervention including possible hospitalization

So it is fairly well established that an infection doesn’t mean dire consequences for everybody so a million infections can cause only so many fatalities. On the same note, the sound bite that all of us are going to it doesn’t have that much of a apocalyptic dread around it.

Also, normal hygiene is a good practice and reduces any infections but this virus is supposed to have a much higher replication factor, meaning an infected person may end up infecting up to 3 persons compared to just 1 by the seasonal flu in US.

The mitigation efforts based on quarantines , travel ban, lockdown, stay at place shelter are meant minimize the spread, along with higher and greater adherence to hygiene based on correct guidelines based on the studies conducted.

When you consider these factors and understand that how much each small event can impact the network effect of the spread, it seems quite understandable the wild fluctuations in projections and also gives hope that the harder the effort at each possible inflection point.

The models will ll keep getting adjusted, hopefully projections will keep trending downwards and each life will be fought for with full force of the society.

This is definitely one case when knowing the future will help prevent it. Let’s try to make each forthcoming model as wrong as possible.

Three posts from on the topic

Coronavirus Case Counts Are Meaningless*

Best-Case And Worst-Case Coronavirus Forecasts Are Very Far Apart

Why It’s So Freaking Hard To Make A Good COVID-19 Model 

Leave a Reply

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