CDC Excess deaths associated with Covid 19 report is issued ever week with weekly data. This data is disaggreagted to daily by a “spline” which produces a smoothed data series. The last date is 10/24th. In the past revisions have occurred, but lately the series has had stability with only small revisions.




Note that when the daily adjusted deaths are extrapolated forward they turn down. The extrapolation is CDC excess regressed to smoothed daily. Caution should be applied and that this is very “twitchy”.

The above plot is repeated but with the daily reported in line form so as to show the great day to day differences. The periodic waves is usually weekly with data dropping on weekends and then “catching up” through the weekdays.




## SIR Models

Using the empirical data of reported deaths, a “Susceptible”, “Infected”, and ’Resolved" model is made. Deaths are lagged 18 days - the consensus currently on mean time from infection to death. Then an Infection Fatality Rate of .5% is used, the consensus mean as to how many infected die. The IFR can have a wide range from .1% to 1% given the IFR is changes given the age cohort with most studies seeing IFR at a low value and then moving quickly up at age 70 or older moving to 20% for over 80 years.

The number of susceptible is seeded, the “S” in the model, with a starting point of 30% of the Texas population based on past geographies experience whcich have ranged from 14% to 50%. This si loosely called “herd immunity threshold”.

The growth of new infections is derived and assuming the infectious period lasts 12 day, which is the consensus from most severity reports, the “I”, infected, is derived. The “I” is used to derive the “R”, the resolved, and then the time series of the SIR is derived.

To rpeat this is based on reported deaths. Below the following the deaths adjusted for CDC Excess Deaths will be used.


The SIR using the reported deaths input is now provided on a ratio basis.


The R(0) for the reported deaths based SIR. R(0) is derived from the “beta”, or how many become infected, and the “gamma” the daily rate of those no longer infected and resolved. Beta to gamma ratio is R(0).



The R(0) derived from reported deaths is plotted with daily reported deaths. There is an obvious correlation once the lag from infection to death is considered.


The SIR is again derived but using adjusted deaths for CDC EXcess deaths. The same 18 days to death and .5% IFR used again.



The Infected from the SIR model above is compared to active cases. Both assume that infectious period is 12 days.

```

The R(0) derived from adjusted deaths is plotted with daily reported deaths. There is an obvious correlation once the lag from infection to death is considered.





Another SIR model is built using a constant R(0) for the entire period. For Texas that is an R of 1.7 which calibrates the lastest adjusted deaths to the model deaths.

The I, or infections, for each SIR is shown.


The R(0) for the three models is provided.




Three cumulative deaths versions are plotted, one the deaths adjusted for CDC Excess, one the reported deaths and the SIR model deaths which uses the unchanging R of 1.7.
Note the large differneces between the three.