Last US pre-election COVID-19 spread assessment
Seven days ago we reported here that all of the swing states in the US Presidential election, which is next Tuesday had growing reproduction of the disease and nothing has really changed since then except new daily case numbers have grown as predicted. Here is the US reproduction number (Reff) and new daily case map as at 31 October 2020.
The model shows the overall US Reff=1.3 and new daily cases are now hitting record highs. The second wave is already worse than the first wave in terms of the number of infected people. And it will get worse in the next two weeks.
Before we do a final pre-election round up of swing states, first a very general oversimplified primer on what we can and can't predict and the relationship between the reproduction number (Reff) and new daily cases. Skip below if you want to see the swing states.
A recap on the CloseAssociate modeller and making predictions
The CloseAssociate model uses a novel numerical method we developed in collaboration with University of Illinois, Urbana Champaign. The technical paper is here. We discovered you could use today's and previous daily case numbers and use numerical methods to get an accurate assessment of today's Reff. This allows us to provide short term predictions of daily cases for the next 10-14 days.
The reproduction number tells you how many people an infected person will infect. If Reff is 2.0 then each infected person is currently infecting two others. That dynamic drives the exponential nature of pandemics. The 10-14 day prediction is in part a function of the nature of the lag periods in COVID-19 function. With COVID-19, epidemiologists have learned and are publishing papers on the course of the infection and disease. It is clear that there is an average 5 day lag between infection and the onset of symptoms. Many remain asymptomatic and never show symptoms but are still potentially infectious. From about the third day after infection people start to shed the virus and infect others. Then there is the lag between the onset of symptoms and the realisation that you should seek medical help and a test. And, then there is a lag between getting the test; a lag in receiving the result; and a lag between receiving the result and going into isolation. These lags gives us confidence to predict that if Reff is greater than 1.0 then there will be more new daily cases within 10-14 days. There are other factors at play, but this will have to do as a rough explanation.
Beyond two weeks it becomes very difficult to predict because we are into the realms of human behavior. Governments set rules but people also determine their own behaviors. So, for example, if parents perceive too much risk and are fearful they may take it upon themselves to withdraw their children from school even if the government mandates the school stays open. On the other side of the coin, people may get tired of lockdowns and defy their government's orders. How they behave is a function of things you cannot model.
But back to the US swing states.
Wisconsin's first wave
Wisconsin's Reff has fallen appreciably to just over 1.0. A week ago it had Reff=1.8. However new daily cases are appreciably higher. For Wisconsin, this is really the State's first wave and on 29 October a record 4,869 new daily cases was logged. If Reff stays at 1.0 or dips lower we expect to see new daily cases to plateau or fall over the next two weeks. But in the short term we can expect the Cheese state to have growing case numbers.
Michigan second wave set to grow
Michigan's Reff=1.5 and hasn't come down in the last week. On 29 October its new daily cases were above 4,000. If nothing changes, expect new daily cases to grow by 50% or so in the next two weeks. The second wave is much bigger than the April first wave peak.
Ohio's current wave continues to build
Ohio's Reff=1.5 which is what it was last week. Its new daily case numbers are climbing steeply and can be expected to grow. This current wave has been building since mid-September 2020 and is arguably the state's third wave.
Arizona's second wave not yet as bad as July
Arizona's Reff=1.4, virtually the same as last week but daily case numbers grew from 890 to 1315 which is in line with our short term prediction. Arizona's first wave peak of nearly 5,000 was much worse but unless Reff declines all bets are off.
North Carolina's second wave continues to grow
North Carolina's Reff has moderated from 1.4 to 1.3 in the last week and daily cases have grown from 2584 on 24 October to 2885 on 29 October which are record numbers for the state.
Pennsylvania's Reff continues to grow
Last week our model calculated Pennsyvania's Reff=1.3 and this week Reff=1.4. The second wave is already worse than the early May first wave peak. In the next 10 days we can expect COVID-19 infections to continue to set new state records for daily cases. Pennsylvania is reported to be a pivotal swing state next Tuesday.
Florida's has a growing second wave
Last week, the CloseAssociate model's algorithm calculated Florida's Reff=1.2. One week on Reff=1.3 and cases have grown consistently above 4,000 cases per day. A look at the county map shows consistent infectious spread of the disease across the whole State.
Tomorrow we look at the situation in Europe which continues to worsen by the day, and also look at some of the more populous US states including California and the Northeast megalopolis states where there seems to have been recent upturns in their reproductions numbers.