Showing posts with label polynomial models. Show all posts
Showing posts with label polynomial models. Show all posts

Monday, March 16, 2020

Update -- COVID-19 Model: Estimated Number of Infected Individuals Based on the Number of Deaths

I've had a chance to examine my model in light of evidence from South Korea that I collected from the Johns Hopkins CoronaVirus Resource Center (https://coronavirus.jhu.edu/map.html.) South Korea has clearly had one of the best if not the best COVID-19 mitigation strategy that has a centerpiece aggressive testing. Therefore, South Korea's numbers should be the most up to date and accurate.

As of 16 March 2020, South Korea's numbers are:

  • 8,162 infected
  • 75 deaths
Johns Hopkins doesn't breakdown the deaths by age group, however, I entered what I believe are realistic numbers into the fields. Here are the results:


Age Ranges, YearsRate of DeathReported DeathsProjected Number of Infected at Present (based on average time of illness until death)Projected number of who are infected nowProjected number of deaths based on the number currently infected (based on average time of illness until death)
80 plus0.14821142803119
70-790.0817213120296
60-690.03613361204374
50-590.0138615348145
40-490.00471750989940
30-390.002420001131423
20-290.002420001131423
10-190.002150028286
0-90.0094853600






Infection rate2.0




Average time of illness until death (weeks)2.5




Total Infected based on current number of deaths8528




Total Infected Now48244




Total number of deaths75




Projected total number of deaths based on the current number of infected individuals424






The numbers to focus on are "Total Infected based on current number of deaths" and the "Total number of deaths." I'm assuming that the total number of infected in South Korea is close to being complete. I'm estimating that South Korea has 8528 infected individuals. However, because of the lack of testing in the US, I would estimate that when the US reaches 75 deaths, that in reality there are 48244 infected. And that the US will see 424 deaths within 2.5 weeks.

I noted one flaw in my estimator. Every cell of reported deaths from 80 plus to 10-19 must have a number other than zero to achieve a reasonable number of infected estimate for that age group. I'll see if I can build in correction mechanism that would enable one to enter zero for an age group and still have a reasonable estimate of the number who are infected. 


Update: Estimate of the Number of Infected in the US Based on Current Number of Deaths (16 March 2020)


As of today, the number of COVID-19 deaths is 74. Based on that information, here's what I estimate to be the number of infected.


Age Ranges, YearsRate of DeathReported DeathsProjected Number of Infected at Present (based on average time of illness until death)Projected number of who are infected nowProjected number of deaths based on the number currently infected (based on average time of illness until death)
80 plus0.14821142803119
70-790.0817213120296
60-690.03613361204374
50-590.0138615348145
40-490.00471750989940
30-390.002420001131423
20-290.00231500848517
10-190.002150028286
0-90.0088550070






Infection rate2.0




Average time of illness until death (weeks)2.5




Total Infected based on current number of deaths7966




Total Infected Now45062




Total number of deaths74




Projected total number of deaths based on the current number of infected individuals419





I decided to use fundamentally the same distribution pattern as I estimated for South Korea. These are the results:


  • The estimate for the number infected throughout the US this day is 45,062. 
  • 7966 people were infected 2.5 weeks ago.
  • Approximately 2.5 weeks from today (end of March, first of April) I estimate 419 people will have died from COVID-19. I suspect that's a low estimate. If 17% of those infected require hospitalization at one point or another, I suspect that many will not receive treatment because of lack of resources and personnel. And many more will die due only to the lack of treatment because of the lack of preparation. 


A Second Update Regarding the US (16 March 2020)


I just heard that the number of deaths is 80. So based on that new number, these are my estimates:

Age Ranges, YearsRate of DeathReported DeathsProjected Number of Infected at Present (based on average time of illness until death)Projected number of who are infected nowProjected number of deaths based on the number currently infected (based on average time of illness until death)
80 plus0.14824162917136
70-790.08182251273102
60-690.03613361204374
50-590.01310769435157
40-490.00471750989940
30-390.002420001131423
20-290.00231500848517
10-190.002150028286
0-90.0090851390






Infection rate2.0




Average time of illness until death (weeks)2.5




Total Infected based on current number of deaths8176




Total Infected Now46250




Total number of deaths80




Projected total number of deaths based on the current number of infected individuals453




  • The estimate for the number infected throughout the US this day is 46,250.
  • 8176 people were infected 2.5 weeks ago.
  • Approximately 2.5 weeks from today (end of March, first of April) I estimate 453 people will have died from COVID-19. 

Saturday, March 14, 2020

COVID-19 Model: Estimated Number of Infected Individuals Based on the Number of Deaths

Background


I hope this post might do the people in the US some good. I created a model to estimate the number of currently infected individuals based on the number of people who have died from COVID-19. 

My motivation for creating the model I'll discuss below originates from my feelings about the absolutely abysmal and utterly incompetent response by the Trump Administration to addressing the threat that COVID-19 posed. 

The Trump Administration had clear warnings that this pandemic was coming. In fact it's being reported that HHS Sec. Azar became alarmed weeks ago about the disease wave that was about to hit the US and Trump would not listen or direct anyone to do anything  related to preparing for the coming pandemic. I'll state here and now that this lack of response, this terrible failure is nothing short of criminal. If Trump were a military officer, he would be prosecuted for dereliction of duty. He certainly has failed the country that he took an oath to protect. 

And Trump has gall to tell everyone that he's not responsible. Or that the actions and policies that he implemented were actually implemented by someone else. Telling lies, upon lies, upon lies is the Trump way. Also Trump constantly states that nothing is his fault. Trump acts as if he's the victim and if you know Trump, that's his typical response. And unfortunately, it's a response that has gotten him out of a lot of trouble in his lying, miserable, criminal and poorly managed life. Trump is responsible, the "buck" stops with the President. All this should never be forgotten. 

As a result of Trump and rest of his sycophantic horde's inaction, the US finds itself in a terrible position with respect to understanding the progress of the spread of COVID-19 because of the failure to develop the capability for mass testing for COVID-19.  The kind of testing program that other countries found the wherewithal to develop, but the Trump Administration seems unable to do.

In response, I realized that there might be a way to estimate the current number of infected individuals based on the number of people who have died from COVID-19. 

The Model


The model is partly based on the Report of the WHO-China Joint Mission, 28 February 2020 by the World Health Organization. (https://www.who.int/docs/default-source/coronaviruse/who-china-joint-mission-on-covid-19-final-report.pdf) The study is based on 55,924 laboratory confirmed cases. What interested me from the study was death rate based on the age of the infected individuals. That information is shown below:

Age Ranges, YearsRate of Death
80 plus0.148
70-790.08
60-690.036
50-590.013
40-490.004
30-390.002
20-290.002
10-190.002
0-90.0

Using the information shown above allowed me to create the model you see below. I have included "dummy data" show how it operates:

Age Ranges, YearsRate of DeathReported DeathsProjected Number of Infected at Present (based on average time of illness until death)Projected number of who are infected nowProjected number of deaths based on the number currently infected (based on average time of illness until death)
80 plus0.148200135176441131
70-790.08506253536283
60-690.036256943928141
50-590.013151154652785
40-490.00461500848534
30-390.002420001131423
20-290.00221000565711
10-190.002150028286
0-90.00110362400






Infection rate2.0




Average time of illness until death (weeks)2.5




Total Infected based on current number of deaths9928




Total Infected Now56160




Total number of deaths303




Projected total number of deaths based on the current number of infected individuals1714






The number of Reported Deaths (within the age range) is variable. The number of deaths is the input. It's a simple calculation to determine the number of people who were infected when those who died were infected. However, dying from COVID-19 is not like dying from 1918-1919 influenza where you could come down with it in the morning and be dead by the end of the day. It's my understanding that it takes about on average 17 days from onset of symptoms until the person dies. That's about 2.5 weeks as shown in the model. However, in my model that is a changeable parameter. What is also changeable is the infection rate. That is, the number of people an individual will infect. In this example, I have set it to 2.0 because indications have been that the number of infected doubles each week.

Note that the death rate for children 0 to 9 is zero. There have been no reported deaths for children in this age range. That has necessitated some data fudging on my part to get what would seem to be a realistic estimate of the actual number of infected individuals. Children may not die from the virus but they are extremely effective carriers and spreaders, and need to be considered in a model that needs to estimate the number of infected individuals.

So, using the data that I have input in each age range cell (the number of reported deaths by age range), I have estimates for the number of individuals who were infected when those who died were infected. And the number of infected individuals should would be expected to be infected currently. I also estimate the number of deaths that will occur in the future, in this case 2.5 weeks from this point in time. 

One more thing, the number of infected individuals for those 0-9 is based on the average number of infected individuals from 10-19 to 80 plus. Since I'm not an epidemiologist, I considered that the average of all other groups would be the best estimate for the 0-9 age group.


For those interested in receiving a copy of this model

I created this model using Apple's Number spreadsheet. I can convert the spreadsheet to MS Excel. If you are interested in receiving a copy of this model, send me an email. My contact email address can be found at the end of "About Me." 

This is a completely new model. And I'll likely update it once more test data become available. I'll publish additional articles when I have updated and revised the model.

One more thing to note: if you use this model, I suggest inputing the total number of deaths from COVID-19.  I'll have more to say about this issue in future articles.