Saturday, May 23, 2020

Who Wants To Predict The Future? COVID-19 Data Forecasting for Beginners

Let's get back to some basic math. 

Here is the data from May 22, 2020 as posted on the Florida COVID-19 Website.

It reveals some interesting information.

To begin with, the mortality rate of COVID-19 for the State of Florida is 4.4% (2,190/49,451 X 100 - divide the number of deaths by the number of positive cases and multiply by 100).

The mortality rate for Palm Beach County is 6.1% (311/5,072 X 100). This is the HIGHEST rate for any county in Florida. Yet it has been less than a week since the "defcon" rate for this county was upgraded.

So how can you prognosticate? Start by the realization that we have become a society so self-centered and focused on the fulfillment of personal desires that the responsibility for the greater good means nothing, and is often considered an inconvenience.

Combine that with the mobility rate practiced by citizens in their quest for self-enjoyment, and you have a recipe for spread and spike in the Coronavirus in Florida.

The county immediately to the north has repeatedly had to put in place "protective" guidelines (including closing boat ramps and requiring id for use of public beaches) as a result of the refusal of Palm Beach County residents to take reasonable protective measures.

Yet rather than behave responsibly, those residents continue to ignore reasonable guidelines (masks, social distancing, basic restraint) and impose their indifference to risk on others.

We have become an "immediate gratification" society, and if it doesn't happen within minutes of our expectation, we get angry. Look at the success of Amazon Prime as an example - "Next Day Delivery" has fueled an upsurge in the "I want it now!" syndrome.

It should come as no surprise that when there isn't an immediate upsurge in infection rates or deaths, people become even more lackadaisical in their precautions.

Let's look at some simple prediction guidelines. 

In terms of time frames, there are no absolutes - the time from exposure to symptoms has been given as a range of two days to two weeks. And the range from exposure to death can be three to five weeks. 

But these aren't absolutes either, just an approximation given a "perfect world" and laboratory conditions, as well as anecdotal evidence.

So now let's look at some Florida metrics. Notice how there is a subtle manipulation of the perception right out of the gate. 

The data in this histogram below doesn't show ALL of the new Florida cases, just those for individuals identified as Florida Residents. Yet there are many people in Florida that have residency elsewhere.

What does that mean in practical terms? The actual number of new cases in Florida is far higher than that shown when you add in the "Non Florida Residents".

Now look at the high data point on the left of the above graph - on April 23 the number was 1.2K, so by extrapolation, there should be a "spike" in deaths anywhere from one to three weeks later, give or take.

Voila, on May 4 - a "spike" in the data on the graph. And once again, keep in mind that this data is ONLY those individuals identified as Florida Residents.

Given the recent termination of the data chief for the Florida COVID-19 website I found it quite intriguing to see a "glitch" in the histograms for the Resident Deaths.

Above is the data posted on May 23, 2020, and below is the data from May 22, 2020.

In the graph from May 23, the death data reported for May 22, 2020 is 2, and for May 21,  18.

Yet ironically enough, the reporting on May 22 had the number of deaths on May 21 set at 2.  Somehow, 16 more deaths on May 21 "magically" appeared in 24 hours.

Once again, bear in mind that these numbers are only Florida Residents, not the data for deaths of people in Florida who are not identified as residents.

So how can you "predict the future"? Look once again at the histogram below.

There are two more "spikes" in the data as of yesterday. On May 16 there were 1.1K new cases, and on May 20, 1.2K new cases. (In case you aren't sure - K is shorthand for 1000, so 1,100 and 1,200 respectively.)

Given extrapolation, that means around today (one week after the May 16 spike) there should be a spike in deaths that could last a week or more. There should be a second spike beginning on or around May 27.

If you look at a calendar, that means there is going to be an overlap of those spikes in death rates. After all, the above two spikes were only four days apart.

Ironically, the first spike occurred two weeks after Florida entered it's Phase One reopening. This precipitated many individuals ignoring the ongoing request by the governor to continue to practice social distancing and wear masks.

And this is where "flatten the curve" comes into play. It is intended to reduce the strain on the healthcare system that will be caused by wave after wave of overlapping surges in hospitalizations based on these spikes.

What is the solution? 

Recognize that "There are no rights without responsibilities," means that although you have the right to freedom, you have the responsibility to not endanger your fellow humans with your flippant behaviors.

Mask up, keep your distance, be responsible.

No comments: