The data driving the
analysis
This analysis relies on public data sources for cases and
deaths in both countries (OurWorldInData.com) and states
(worldometer.info). In addition, I use news services for
analysis of the latest studies on antibody tests.
Finally, I use cell phone data originally sourced from Google
to measure community isolation (covid19.healthdata.org).

As the pandemic has spread from China, Europe and then
US became the epicenters. By mid March, the news from Italy
and New York, had made both the general population and
the government aware that stopping the exponential growth
was essential and the US was in  an almost universal lock
down.  The idea of “flattening the curve” had become the
universal watchword and several sites were started to collect
and distribute quality infection data. It took a month for the
isolation to stall the infection. By late April, the US infection
had peaked and was showing a clear decline as shown to
the right.












Pandemics occur because infections are essentially a chain
reaction, that naturally produces exponential growth and the
best way to analyze exponentials is using log plots. The
daily  New Case and Death data for the US, shows one
exponent at the  start of the infection before isolation,  and
different exponent after isolation has stalled the infection.
The peak occurs during the transition between open  and  
isolation. It is common to use a 7 day rolling average to
smooth daily and weekly periodic variation.







The international version of the log plot highlights the huge
variation in the progression of the infection in different
countries. The start of the infections seems to have similar
slopes. The differences appear in isolation. New Zealand
lock down early and stopped the infection before it got
started. The US locked down late and has just controlled the
infection. Brazil has slowed but not stopped the growth. In
between, countries had varying success indicated by the
slope post isolation. The idea of  open infection rate, an
isolation infection rate and a transition as the isolation gets
established, are central to  the model of the infection
described in the next section.

It is important to recognize that there is significant
uncertainty in identifying cases and causes of death,
reliability and honesty of data reporting.

Since the start, one of the challenges is that mild infection
could be easily confused  with the flu or a common cold.
There was a huge push to get a reliable test for the virus,
and a test for the antibody that would show evidence that
someone had been exposed and had recovered.
By May 4th the results of the  first large scale antibody
survey were in.

New York State Governor Cuomo said that
"preliminary
findings from an antibody study conducted on 3,000 people
at grocery stores across New York State found a 13.9% had
coronavirus antibodies, suggesting a 13.9% actual infection
rate statewide (21.2% in New York City), which translates to
an estimate of about 2,700,000 actual cases in New York
State (10 times more than the about 270,000 cases that
have been detected and reported officially). Governor
Cuomo acknowledged that the official count reported by New
York State (which still is not including probable deaths as
recommended by the new CDC guidelines) of about 15,500
deaths is "not accurate" as it doesn't account for stay at
home deaths. Based on Worldometer's count (which
includes probable deaths reported by New York City) of
about 21,000 deaths and the 2,700,000 case estimate from
the new antibody study, the actual case fatality rate in New
York State could be at around 0.78%.
"https://www.
nbcnewyork.com/news/local/new-york-virus-deaths-top-15k-
cuomo-expected-to-detail-plan-to-fight-nursing-home-
outbreaks/2386556/

These results were confirmed by a study from Spain with
7,000 people published in mid May. It is clear that there are
10x as many people who have been infected with no
symptoms (“Asymptomatic”), as there are new cases who are
Symptomatic. This is really critical because it affects the final
fatality of the infection and the onset of herd immunity. It also
highlights a huge challenge in identifying and quarantining
infectious individuals. The only way to find these people is
blanket testing of everyone for the virus, independent of
symptoms. It is probably a big reason the pandemic got
established and resisted early containment.worldometer.info