Btw on testing ... this could be a game changer. Point of Care testing (can be in many different offices ... I wouldn’t be shocked if every university gets some of these machines at some point).
COVID-19 Statistical Prediction Model
- Thread starter BoilerBulldog
- Start date
Nobody thinks it can happen here, until it does.
Vigo County reporring it's first death. Person in their mid 50's with a lot of past medical issues.
Army Core of Engineers-Right, they have identified something like 120 sites to develop into a hospital. They are currently looking for contractors to help build them. These places as others have linked can house hundreds or thousands.
I was talking more along the lines of a mobile military type unit hospital. Sure, one is only talking a few dozen per hospital but that is better than turning people away or not having the space.
Asked my fiance about that. She said those doctors are already busy working at military base hospitals trying to treat it there.
The projections here with ranges seem to be very useful, showing beds, icu and ventilators. Hope we end up at low end of range which is huge.
Looks like 110k tests per day is our capacity right now. Good news is new cases slightly down vs yesterday.
Explain Sweden. They were just limiting mitigation to just prohibiting large gatherings and all schools were open. As were businesses. Maybe this changed in the last couple days.
Dr Birx addressed the Sweden question a couple days ago. She said Sweden doesn't have a lot of int'l visitors this time of the year. Compare that to that amount of people coming into the U.S.
In addition imo its tough to compare a country of around 10M to a country of around 300M with 50 independent states.
California has completely derailed any credibility this guy might have. His model predicted california to be on the downside at this point. This weekend has seen a continued surge of cases, well above the peak of his model and accelerating instead of declining. We're also seeing more fatalities than his model showed, already we're over his predicted total.
Again, he is trying to build models with garbage data, and his models are garbage because of it.
Again, he is trying to build models with garbage data, and his models are garbage because of it.
This guy’s conclusions are WAY off. Up until a few days ago he was projecting 1,400 total deaths in the U.S. That number was laughable from the start. Then a couple of days ago he doubled that projection, and that number will end up being way off too.
Bingo.
Yes his model is going to be way off. At the same time the average in 18 medical expert models was 200K deaths, with the most extreme being 1.2M. It is still possible he will be more accurate than the expert models. Everybody was dealing with garbage data.
I don’t doubt that his total case number and total death number will be off some. The reason I keep following his updates is because he is pretty certain that this thing is going to peak by tomorrow nationally, with some states peaking in the next few days. Some states already have according to his models. I think we can all agree that if we wake up Tuesday and the whole country is past the peak, then that’s positive news.
Last edited:
Yeah. The guy is sticking to his guns of 3/30 peak. We’ll see. I do find myself checking his FB page a few times a day to see if he’s updated his models.
He called California peaking Friday. California continues to accelerate. He’s wrong. Game over.
Is it accelerating because of more cases or more tests? People keep saying Illinois is accelerating but they are increasing the number of testing locations and subsequently the number of tests. But what’s not accelerating in Illinois is the positive rate. Since IDPH has posted results the positive hits have been between 13-15%. So on any given day of a 100 people who pass the requirement to be tested 85-87 have the flu or another virus. So as I have contended since the start, its illogical to think a city with 1000s of Chinese nationalists students and many more Chinese visitors or Americans retuning home from travel on one of a multiple of big body jets originating in China landing at Ohare this thing has been running through Chicago since early January. Only difference is you are looking for it.
You are right about him missing on California, and he didn’t offer a great explanation other than saying it must have been a secondary infection of some kind. However, he is right on track with most states. I don’t have anything better to do, and I’m intrigued by all of the models that he does have correct. If you have someone better for me to follow in between the 20,000-2,000,000 deaths predictions that I’ve seen, let me know.
Try to keep in mind that this is just a simple mathematician and teacher in Ohio doing this. He's not making press releases or calling people liars and he isn't seeking media approval. It started as a simple math lesson. He made some initial projections based on the data he had. His initial projection was around 1400 deaths, which was clearly low. He's adjusted that up to just under 6,000 now. Yesterday he said he had 2 clear errors in his original hypotheses. He used the death rate of around 1% but that was running at almost 2% of known cases (not to say that the death rate is actually 2% but it is closer to this for known cases). That doubled his estimate. Also, he was treating the entire country as a whole. New York clearly blew that up. So now he has it by state. He constantly harps on the scientific method. Make hypotheses and use data to prove or disprove. In the end, if he's right or if he's 10 million off, he's just trying to learn and taking people like me on the journey with him.
He doesn't and he hasn't taken a moment to actually read what the guy has been doing all along. Makes a false exclamation that he is trying to predict what will happen and then says he's wrong...game over.
Comes off very insecure honestly.
In terms of whether his model is accurate, it doesn't matter. But this highlights why his models are wrong, he doesn't have enough data to be doing this.
Then no one should be doing it because he’s using the same data as everyone else.
He did revise his California models about 13 hours ago. Here they are and his explanation FWIW. I don't know if his numbers are good or not. California doesn't appear to do a daily update and their numbers get update constantly. It's hard to figure out a good cutoff.
He says:
Ok. Something good happened (I think). California revised it's reported numbers. Normally, I hate this because then I have to back and reconcile everything.
But last time I reported on Calif I was bummed because there was an obvious inflection in the the deaths curve, which typically means secondary infection.
But, as it turns out... they revised the official record... and voila! No inflection point. (I'm happy for my mama!) There remains a glitch in the data, but the rest follows as we normally expect.
I give you two graphs, one smoothed and one not. Because of the change in data, I also included a fine tuning of the parameters, and I show you a smoothed plot for that.
I also include a reference graph/citation from Wiki.
I'ma perta' happy right'a now!
He says:
Ok. Something good happened (I think). California revised it's reported numbers. Normally, I hate this because then I have to back and reconcile everything.
But last time I reported on Calif I was bummed because there was an obvious inflection in the the deaths curve, which typically means secondary infection.
But, as it turns out... they revised the official record... and voila! No inflection point. (I'm happy for my mama!) There remains a glitch in the data, but the rest follows as we normally expect.
I give you two graphs, one smoothed and one not. Because of the change in data, I also included a fine tuning of the parameters, and I show you a smoothed plot for that.
I also include a reference graph/citation from Wiki.
I'ma perta' happy right'a now!
I have read what he's written. He should be commended for at least being transparent about the fact that his model is wrong. That doesn't make him less wrong. He's made a key mistake, and that's trying to build a predictive model when the data is not clean and he doesn't have enough actual good data to do so. That's what has happened. You may not like that I and others have enough experience in this to know, but given that I've been doing this since I left Purdue 20 years ago and others have done similar things, we can see what he's done wrong.
It doesn't mean it's not interesting in a way, but he isn't going to be anywhere close to being right.
Ah, so he's papering over how wrong he was by "revising" his predictions as he goes along. That's....something.He did revise his California models about 13 hours ago. Here they are and his explanation FWIW. I don't know if his numbers are good or not. California doesn't appear to do a daily update and their numbers get update constantly. It's hard to figure out a good cutoff.
He says:
Ok. Something good happened (I think). California revised it's reported numbers. Normally, I hate this because then I have to back and reconcile everything.
But last time I reported on Calif I was bummed because there was an obvious inflection in the the deaths curve, which typically means secondary infection.
But, as it turns out... they revised the official record... and voila! No inflection point. (I'm happy for my mama!) There remains a glitch in the data, but the rest follows as we normally expect.
I give you two graphs, one smoothed and one not. Because of the change in data, I also included a fine tuning of the parameters, and I show you a smoothed plot for that.
I also include a reference graph/citation from Wiki.
I'ma perta' happy right'a now!
Remember this is what he was saying. He was wrong.
Okay, that's fine. For me, I take it as he started with a swag knowing he would have to adjust as data got better and actuals came through.
In general he is sticking to his guns that the peak will be sooner than most thought and deaths will be lower than most thought (not predicting secondary outbreaks at this point).
If we don't see peak in 2-4 days that's probably my first gate where I'd say he really botched it all. Then once deaths exceed 15 or 20k that'd be next. If he shadowed a massive low/high range of outcomes on his chart like everyone else he would be less open to the trolling accusations but I guess that's not his style.
I'm not knowledgeable to say whether I think he'll be right or not (I hope he is) but he's generally been pretty good so far doing it in a way I'm not seeing anywhere else so I'll keep checking in.
It seems to me he is clinging to his projection of cases and now raising the death rate. He should be doing the reverse. Keeping the death rate at 1% and changing his projection of cases. Bottom line nobody else including Fauci is projecting this thing to peak nationally Tuesday. We only have to wait two more days to know he is wrong on his projection of when the curve bends. I think it is cool he did this but because of the lack of early testing his data was incomplete and this his model was off.He did revise his California models about 13 hours ago. Here they are and his explanation FWIW. I don't know if his numbers are good or not. California doesn't appear to do a daily update and their numbers get update constantly. It's hard to figure out a good cutoff.
He says:
Ok. Something good happened (I think). California revised it's reported numbers. Normally, I hate this because then I have to back and reconcile everything.
But last time I reported on Calif I was bummed because there was an obvious inflection in the the deaths curve, which typically means secondary infection.
But, as it turns out... they revised the official record... and voila! No inflection point. (I'm happy for my mama!) There remains a glitch in the data, but the rest follows as we normally expect.
I give you two graphs, one smoothed and one not. Because of the change in data, I also included a fine tuning of the parameters, and I show you a smoothed plot for that.
I also include a reference graph/citation from Wiki.
I'ma perta' happy right'a now!
Fine let’s concede a bunch of people in chi had it in January. Quit harping on it. No one seems to be arguing with you. Lift the whole curve about 10000 or whatever number you want to account for undiagnosed early cases. What’s important now is what is happening now with cases and deaths and is the CURRENT curve bending now?
Lift it by 10,000? I think we probably can double or triple the number of cases. I think a lot of people have or have had this.
I hope cases are 10x and immunity is building. Just need deaths to stay low.
It absolutely does matter because it means communal immunity. Fewer potential victims. It also matters because it slices the lethality rate. Currently the lethality is artificially high because of of testing bias. Add 10k (which is extremely conservative) and you current lethality rate in Illinois goes from 1.4% to .4%. Less than the common flu.
I would strongly suggest that people follow John Ioannidis., Prof of Epidemiology, Stanford School of Medicine. Basically his point is we are making draconian public decisions based on junk data. Interesting he points out that in any normal years 1000s die of that year’s Coronavirus. We just don’t test for it. I. e. We say grandma got sick and got pneumonia. And I would suggest if you put a death counter on the television 24/7 for opioid deaths (130 per day in 2019) you’d have a different approach to that.
If you disagree. Don’t argue with me. Argue with the Prof of Epidemiology at one of the country’s top medical schools. You know. A “scientist.”
Agreed. We know that hospitalizations are correlated to deaths so hopefully both stay low and within what the system can manage.
But we are seeing hospitals get overrun by this disease regardless of the quality of data so letting it go like the flu is not an option.
Yeah, and people often forget that urgent care, emergency depts, and overall hospitals weren’t sitting empty before this began. There is a standard amount of people who need care for anything from trauma, cancer, stroke, MI, etc. If a community’s healthcare infrastructure system buckles, then the body count across all other events/conditions will also be at risk for a spike.
Exactly. I don't see the national guard and army Corp of engineers building field hospitals for the the regular flu. Or The President ordering GM to make 40,000 ventilators because "grandma got pneumonia". People forget the speed of these infections and the strain on the health system because of the volume of people getting sick all at once. They just want to call it "the flu" and let it go.
I hope you are correct. Fauci said tonight that current model showing 100k to 200k deaths in US. Peak in 2 weeks
