Coronavirus On the Beach

Danger Looms for Cape Town.

Unless you understand mathematics, this is going to sound like Neville Shute's - On the Beach.

First the good news. As I pointed out in my last bulletin, the SA authorites have reset the origin to 27 march. Notice we no longer talk about that "miracle hook" in fact it is now obvious  to even the most positive Capetonians that, in the outbreak  areas,  the lock down is not working. The good news is that some Provinces have worked and the regulations are now  tightening up provincial borders , too late, but at least we have begun. The executive ARE getting the idea of compartments. We have a  model for the release from the lock down, too complicated, but it is a start. ,

Now the bad news. The SIR style epidemiologists have still not understood the veracity of the infection vector. If you have taken the time to study the Dengue model, you will see that the infection vector (misquitos in Dengue)  is  driven by the "essential workers" in COVID. Rather than fearing for their lives or at least appreciating that they are risking the lives of others, essential workers enjoy the status of being essential and the income priveleges that come with that status. While their wages depend on not understanding this logic, they will never get what I am talking about. By the time that our leaders (they are essential themselves) figure out that they are themselves,  the infection vector that causes the spread, it will be too late.

 

Bulletin 30 April 2020 - bulletin30apr.pdf

E-mail me when people leave their comments –

You need to be a member of Copeman Academy to add comments!

Join Copeman Academy

Comments

  • 4616198655?profile=RESIZE_930x

  • Good work on the Wits Covid Page.

    https://www.covid19sa.org/provincial-breakdown

    May I make the following suggestions:

    1) There are now a number of parallel efforts to present COVID 19 data. These are all excellent examples of BI. However, when dealing with exponential data the value of backward-looking analysis has limitations. I do understand that this is easier said than done, but we would all be better off if we focus resources on future-looking scenarios.

    Exponential, non-stationary data is best represented in Log-Linear form (see my Western Cape forecast below)

    I understand that putting the researcher on the block - having to produce results with not enough time for thought, will expose the predicter,  but predictions (with the limits of Keynesian probability) are more valuable than descriptions of the past.

    2) Please publish the raw data that you work with and the source code to any modeling. This will make for a far better spread of information

    3) We are already past the provincial level. The key to dealing with COVID19 is to get ahead of the epidemic. National data is no longer relevant and provincial data is fast becoming obsolete. The optimum modeling units are between 1,000 and 10,000 populations. R zero is not homogenous and varies far too much for an aggregate approach. So far the Western Cape is the best delivery of this regional data (although they also dot no publish full data sets and their data can only be retrieved by scraping.  https://coronavirus.westerncape.gov.za/news/update-coronavirus-pre...

    4) I see from the Wits COVID page that you are referencing The Ebola model. I would be most surprised if this provided any accuracy. Compared to COVID the Death rate for Ebola is is simply far too high and Delta is much too low. In COVID delta is 14 days, most Ebola cases are resolved by then.  By substituting death for Recovery, the Ebola model rapidly collapses into standard SIR.

     Amenaghawon Osemwinyen, Aboubakary Diakhaby (2015). "Mathematical Modelling of the Transmission Dynamics of Ebola Virus". Applied and Computational Mathematics. 4: 313–320.

    5) I have been getting far more robust results using the SIRVEC model (Used on Dengue). In  Dengue a Mosquito is the primary carrier. Now if you consider COVID in an urban situation infecting homo economicus, now generally subject to movement restriction, a simplification is to divide the population into  "locked down" and "carrier." By modeling these two distinct Rzero emerge - with Re evolving at different rates far more robust models are derived.

    6) We should try to focus on discussion of policy (no matter how prickly that may be)

    i) Recognize that the lockdown has only had limited success. Yes it has protected the rural areas. No it has not worked in Urban densities. Our rates of infection POST lockdown are now almost equal to European cities PRIOR to lockdown. As of today, we are no going to make that worse (only reflecting in 14 days rime)

    ii) The "ëssential workers" are the carriers. Noting the failure of our current policies to contain the virus in densely urban areas, continue with their path, and even unlocking further movement is likely to have disastrous consequences.

    ii) because of the heterogeneous nature of R zero in South Africa- One strategy fits all policies are doomed to doom us.

    iv) The actions required to arrest Coranvirus  CANNOT be popular. Our politicians are playing to the cameras and playing with our lives. We were most fortunate to have moved quickly on 26 march to lock down international borders. Our policies, post that action, have proved just as spectacularly unsuccessful. Our COVID  honeymoon is over.

    South Africa Provincial Breakdown | Covid-19 South Africa
    Provincial breakdown of coronavirus in South Africa.
This reply was deleted.