Posted on 09/15/2014 2:50:13 PM PDT by scouter
My User Name on Free Republic is Scouter. I have been a member of Free Republic for 14 years. I don't write many vanity posts, but I consider this one to be very important. I had been working on this post for several days, and I was planning to post it tomorrow. But the Drudge Report headline CDC: PREPARE FOR EBOLA has moved up my timeline.
I have developed a model for making future projections of the number of Ebola cases. I have undertaken this project for several reasons. First, out of simple professional curiosity. Second, I believe the time has come to be concerned and to prepare for the possibility that the Ebola epidemic could spread to other countries, including the United States. And third, my daughter will soon begin working as a nurse in a major Pediatric Intensive Care Unit, which will likely see some of the first Ebola cases in the United States, should it make an appearance here.
I am not an epidemiologist, and I have no inside knowledge about the current Ebola epidemic. But I have spent the last 26 years of my career applying computers to the practice of medicine and to medical data. I hold a Master's Degree in Medical Informatics from a major university known for their expertise in that field. I currently work in that field at a large, famous, metropolitan teaching hospital. I am remaining anonymous only because I don't want my employer to be held responsible for this post in any way. It is my work exclusively, and I am responsible for any information or projections it makes.
The numbers produced by this model are "projections", not "predictions". That is to say, I do not predict that there will be x number of Ebola cases on any given future date. Rather, I "project" into the future, assuming a constant Daily Transmission Rate (DTR), based on past data. Any number of factors can influence future DTR, in either a positive (bad) direction, or in a negative (good) direction. There is no way to know how these factors will actually play out. If there were, then we would be able to make actual preditions. As it is, we are left only with the ability to say "If Ebola continues to spread at the same rate it has been spreading for the past x number of days (or months), then this is approximately how many people who will have contracted the disease as of this particular date in the future." Not ideal, for sure, but still quite useful to understand the seriousness of the situation.
I have validated the model based on actual data by calculating the DTR for various periods of time and comparing the model's projections with what actually happened in subsequent periods. This is the same concept that is being used by epidemiologists at CDC and elsewhere. It is a valid method, within the constraints I have mentioned above. My model has been completely in line with projections I have seen quoted in the mainstream news. It works quite well. If anything, my model's projections are a bit more conservative than some projections you may have seen in the mainstream media. I just take them out further than you have seen in other places.
That being said, the following projections are based on the Daily Transmission Rate (DTR) from June 1 through September 10, the last date for which I have data. The DTR has remained relatively stable over that period. To be conservative I assumed that the reported number of cases represent the true size of the epidemic. However, the WHO, CDC, Medicins Sans Frontieres, and Samaritan's Purse all agree that the number of reported cases represents only 25% to 50% of the true number of cases. I have decided to be conservative in the numbers published below, but the model allows you to adjust this percentage.
As you review these projections, remember to pray for all those who are currently affected by this terrible disease, those who have it, those who will die, and their families. Do not forget that these are real people with eternal souls, who will either go to heaven or to hell, depending on whether or not they die in friendship with God. Pray, too, for an end to this epidemic. Do not underestimate the power of prayer!
The following projections assume that the currently reported cases represent 100% of the true epidemic size. In other words, that there are no cases that were missed by the epidemiologists. We know this not to be true, so we know that the "best case" is something worse than this, assuming the Daily Transmission Rate remains stable.
Scouter Ebola Projection Model Version 1.0 - Ebola Case Projections
*********************************************************
Projection Parameters
*********************************************************
Start Date: 6/1/2014
End Date: 9/10/2014
Reported cases represent 100% of the true epidemic size
Daily Transmission Rate (DTR): 1.00422415489918
*********************************************************
Weekly for the Next 8 Weeks
Date Cases Deaths Daily New Cases Daily New Deaths
========== ==================== ==================== ==================== ====================
09/10/2014 4,845 2,376 171 84
09/17/2014 6,227 3,054 219 108
09/24/2014 8,003 3,925 282 138
10/01/2014 10,285 5,044 362 178
10/08/2014 13,218 6,482 465 228
10/15/2014 16,988 8,331 598 293
10/22/2014 21,833 10,707 769 377
10/29/2014 28,060 13,761 988 485
End of Month for the Next Year from the End Date
Date Cases Deaths Daily New Cases Daily New Deaths
========== ==================== ==================== ==================== ====================
09/30/2014 9,923 4,866 349 171
10/31/2014 30,146 14,783 1,061 521
11/30/2014 88,357 43,331 3,111 1,526
12/31/2014 268,427 131,637 9,451 4,635
01/31/2015 815,475 399,911 28,713 14,081
02/28/2015 2,224,815 1,091,055 78,336 38,416
03/31/2015 6,758,941 3,314,601 237,983 116,707
04/30/2015 19,810,535 9,715,135 697,531 342,071
05/31/2015 60,183,993 29,514,379 2,119,084 1,039,204
06/30/2015 176,399,989 86,506,991 6,211,061 3,045,920
07/31/2015 535,899,508 262,806,446 18,869,075 9,253,441
08/31/2015 1,628,051,594 798,400,534 57,323,860 28,111,763
09/10/2015 2,329,918,242 1,142,597,677 82,036,655 40,230,979
The following projections assume that the currently reported cases represent 75% of the true epidemic size. Remember that Medicins Sans Frontieres, Samaritan's Purse, the CDC, and WHO all agree that the number of reported cases already vastly underestimates the true size of the epidemic. They say by a factor of 2 to 4.
Scouter Ebola Projection Model Version 1.0 - Ebola Case Projections
*********************************************************
Projection Parameters
*********************************************************
Start Date: 6/1/2014
End Date: 9/10/2014
Reported cases represent 75% of the true epidemic size
Daily Transmission Rate (DTR): 1.00422415489918
*********************************************************
Weekly for the Next 8 Weeks
Date Cases Deaths Daily New Cases Daily New Deaths
========== ==================== ==================== ==================== ====================
09/10/2014 6,460 2,376 235 115
09/17/2014 8,373 4,106 305 149
09/24/2014 10,853 5,322 395 194
10/01/2014 14,068 6,899 512 251
10/08/2014 18,234 8,942 663 325
10/15/2014 23,635 11,591 860 422
10/22/2014 30,635 15,024 1,115 547
10/29/2014 39,709 19,473 1,445 708
End of Month for the Next Year from the End Date
Date Cases Deaths Daily New Cases Daily New Deaths
========== ==================== ==================== ==================== ====================
09/30/2014 13,556 6,648 493 242
10/31/2014 42,764 20,972 1,556 763
11/30/2014 129,996 63,750 4,729 2,319
12/31/2014 410,085 201,107 14,920 7,317
01/31/2015 1,293,657 634,413 47,066 23,081
02/28/2015 3,651,570 1,790,739 132,851 65,150
03/31/2015 11,519,271 5,649,079 419,092 205,524
04/30/2015 35,016,714 17,172,283 1,273,972 624,759
05/31/2015 110,464,001 54,171,820 4,018,881 1,970,869
06/30/2015 335,792,614 164,673,529 12,216,744 5,991,122
07/31/2015 1,059,294,023 519,480,413 38,539,038 18,899,640
08/31/2015 3,341,657,268 1,638,757,001 121,575,553 59,620,953
09/10/2015 4,840,743,028 2,373,912,370 176,115,013 86,367,239
Obviously, there are many factors that will affect these projections. Rather, this model simply projects the number of cases and fatalities based on the current Daily Transmission Rate (DTR), which has been stable for about 3 months. Consider the following other factors that are likely to change the DTR (either for good or for bad) as we move forward from today:
While the numbers quoted above are grim, they do not yet represent fact. Do not panic, but do not be complacent, either. Any preparations you make to "shelter in place" will serve you well for other contingencies, too.
On the other hand, epidemiologists are already saying that the number of cases is already doubling every two weeks. That means that the numbers I've posted above are actually quite conservative.
This model is contained within a macro-enabled Microsoft Excel 2010 spreadsheet (i.e., a .xlsm file). I would be willing to share it with other Freepers if someone can provide a place to post it for download and can tell me how to sanitize my name from it (again, I don't want my employer to be in any way held accountable for this).
Neither he, nor the smartest woman in the world, have thought that far ahead.
You're right, of course. And I've tried to make that exact point in different ways throughout this thread.
Nevertheless, the point of doing the math for those who won't do it themselves is to show the pooh-poohers that this really is something to be concerned about. Sooner, rather than later.
I've wanted to show the growth of the epidemic, based on actual data, not on vague statements we're hearing from the government such a "the epidemic will get out of control if we don't do something soon." Because that's all we get. And when they do provide projections, it's only until the end of September. Based on current rates, the real action starts in December or January. The numbers help people see that for themselves.
If they see in real numbers where this is heading in the near future, perhaps some people will take heed.
Were you trying to post a link? The image you posted takes you to tinypic.com.
Basically, you are showing what happens with exponential growth.
Although I think your basic point is right, you missed some things that would greatly improve your model. (I used to teach statistics at the University level, and I still provide occasional expert witness work in the field. I’ve also done a bit of work with bio-statistics.)
If you want to have a more accurate model you will need to add a way to separate the dead and the recovered from the currently infectious.
You will also need to include a way to show growth slowing once the pool of uninfected people in a given area has shrunk significantly. Once a given area is wiped out there will be no more growth in cases from that location. A town can only die once.
At this stage these issues do not make a lot of difference in the numbers. But once this thing gets really going they will become major points.
If you want to get really complicated, and really SCARED, then give some thought to what happens when the medical system has been destroyed. How long it is from there until we have enforced movement restrictions. And how long the grid will hold up.
Wasn't trying to post a link. I often use images to notify The Ping List, instead of writing the word, "Ping."
Pretty much, but with a twist. I've been attempting to account for the rate of change in the rate of transmission. Another FReeper has convinced me that my formulas overestimate to some degree. That overestimation becomes more significant the further you move out into the future. I'm working on a new approach, based on more simple concepts like compounded interest. But I still want to account for the change in the rate of transmission that naturally occurs over time.
Although I think your basic point is right, you missed some things that would greatly improve your model... If you want to have a more accurate model you will need to add a way to separate the dead and the recovered from the currently infectious.
Yeah, I mentioned that. But it isn't quite that simple... Are the recovered really immune? Can't they get it again? Are they still contagious for some period of time (seems so, to some degree... the virus is present in the semen of males for up to 6 weeks, I think).
You will also need to include a way to show growth slowing once the pool of uninfected people in a given area has shrunk significantly. Once a given area is wiped out there will be no more growth in cases from that location. A town can only die once.
To further emphasize this point, taken out a few months further than I posted, the number of cases goes into the trillions. Obviously that can't happen. I don't even think it will go into the billions, and perhaps not even into the millions. But at what point will the shrinking pool of uninfected persons start to have an effect? And how quickly? I'm not sure how to incorporate that. I would be open to suggestions.
This is an extremely complicated point that would take a lot of data I don't have available to me to represent. How much the pool has shrunk in Monrovia will be different from how much it has shrunk in Sierra Leone. I may be able to account for it at the country level, or at the macro level (i.e., the world), but any further down that that will be a challenge.
If you want to get really complicated, and really SCARED, then give some thought to what happens when the medical system has been destroyed. How long it is from there until we have enforced movement restrictions. And how long the grid will hold up.
Another great point that's going to require some thought.
Read the link at post 2160 on the surveillance thread.
It’s even more complex than that. A “real world” prediction requires some assumptions about how governments will respond, and how people at the individual level will respond. These responses will vary depending on the local culture and how the disease has progressed in that location.
So... All things considered... Making a model of this is useful from a hypothetical “what if” standpoint, but not from a real world prediction standpoint.
HOWEVER, the public has no clue how bad this could get. Most of the public, including many Freepers, are convinced this no big deal because the numbers currently being cited are not large relative to other diseases. Models such as yours can help overcome their ignorance and apathy.
Which is exactly why I've emphasized over and over that this model projects into the future based on a constant transmission rate. It predicts nothing.
HOWEVER, the public has no clue how bad this could get. Most of the public, including many Freepers, are convinced this no big deal because the numbers currently being cited are not large relative to other diseases. Models such as yours can help overcome their ignorance and apathy.
Which, aside from my own professional curiosity, and the fact that my daughter will be on the front lines if it appears in my area, was my reason for making it and going public with it... to convince the pooh-poohers that it's not something to pooh-pooh. It's time to pay attention and to prepare. If it pans out according to the model's projections you'll be in good shape. If it doesn't, then you'll be prepared for the next big snowstorm. Either way, it's good.
...and schools, and open offices, and public transportation, and pretty much everywhere that groups of people gather.
Don't leave the house for sixty days or so, and the plague may pass you by. There is still time to assemble the necessary supplies.
It will be an obvious (Yes/No) decision by the end of November.
"Never bet on the end of the world. How are you gonna collect if you win?". (attributed to K. Denninger)
bkmk
Bookmark
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.