Yes, they're updating every day and detailing how the models are changing with new information in their write-ups. Their noted margins of error around their curves are huge, so every possible piece of additional data helps to tighten things up. IHME is super well respected in the health analytics world. They do fascinating work.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
Interesting. I just checked Washington and they are saying for us too we are at the peak (where yesterday it said we had another week). They also lowered our total death estimate from close to 1,000 to 632.
Models continually get better as they have more data to feed in to them, so I'm really hoping this is a more accurate picture.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
Interesting. I just checked Washington and they are saying for us too we are at the peak (where yesterday it said we had another week). They also lowered our total death estimate from close to 1,000 to 632.
Models continually get better as they have more data to feed in to them, so I'm really hoping this is a more accurate picture.
California’s death estimate is down to under 1800, it was over 6000 at the last update. Nice to see the numbers going in a good direction.
Post by sillygoosegirl on Apr 6, 2020 0:16:18 GMT -5
Oregon's projected peak has moved more than 2 weeks, with plenty of resources projected. Hopefully they are right.
I've been scratching my head at the projections for Oregon for a while... it was showing 5 weeks to the peak almost 2 weeks after our shelter in place order last I checked, and that didn't seem possible.
Oregon's projected peak has moved more than 2 weeks, with plenty of resources projected. Hopefully they are right.
I've been scratching my head at the projections for Oregon for a while... it was showing 5 weeks to the peak almost 2 weeks after our shelter in place order last I checked, and that didn't seem possible.
CO's data is wildly different than the last update which was Friday, I think. I can't remember the numbers but we were really short on regular beds, ICU beds, and more than short 113 vents. They also have the peak usage date as 4/4 and peak deaths per day as today. And the deaths not increasing after 5/1. Something seems off but data analysis isn't my strong suit.
Yes, they're updating every day and detailing how the models are changing with new information in their write-ups. Their noted margins of error around their curves are huge, so every possible piece of additional data helps to tighten things up. IHME is super well respected in the health analytics world. They do fascinating work.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
I should have read the newest posts. So many other cities/states are so much worse. I don't see people around me social distancing at all. They're meeting up at schools to let their dogs play. (I could tell they didn't live together because one couple got in a car and drove away.) They're doing yoga in the park today. I walked past a gym having a class with the curtains drawn. My neighbors are having parties. The numbers on this site today seem off but I don't actually know.
Yes, they're updating every day and detailing how the models are changing with new information in their write-ups. Their noted margins of error around their curves are huge, so every possible piece of additional data helps to tighten things up. IHME is super well respected in the health analytics world. They do fascinating work.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
To be honest, I’m not sure how much I trust their data. For example, in Michigan they are predicting peak deaths on April 9 (that part I’m not questioning, as it’s consistent with other reports I’ve seen). But then if you look at the graph, they are predicting no more deaths before the start of May. I just don’t see how it’s possible that 3 weeks from now, there will be no more deaths in MI. Even if we get things massively under control, and the hospital systems are not even close to overwhelmed at that point, I have to imagine there will still be deaths then. People are still being exposed today, and could be to the point of being in critical condition 3 weeks from now. Even with the best possible medical care, some percentage of cases will be fatal. And maybe I could be persuaded to think that just maybe deaths would be so few that they just don’t register on the scale of the graph, but their projected total deaths by August doesn’t show an increase at all, either. I would think the cumulative deaths by then would be showing a change in the line at some point after May 1. Even though I really wish we would be done with this by May, it makes me question the rest of their results.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
To be honest, I’m not sure how much I trust their data. For example, in Michigan they are predicting peak deaths on April 9 (that part I’m not questioning, as it’s consistent with other reports I’ve seen). But then if you look at the graph, they are predicting no more deaths before the start of May. I just don’t see how it’s possible that 3 weeks from now, there will be no more deaths in MI. Even if we get things massively under control, and the hospital systems are not even close to overwhelmed at that point, I have to imagine there will still be deaths then. People are still being exposed today, and could be to the point of being in critical condition 3 weeks from now. Even with the best possible medical care, some percentage of cases will be fatal. And maybe I could be persuaded to think that just maybe deaths would be so few that they just don’t register on the scale of the graph, but their projected total deaths by August doesn’t show an increase at all, either. I would think the cumulative deaths by then would be showing a change in the line at some point after May 1. Even though I really wish we would be done with this by May, it makes me question the rest of their results.
I should have added "is the model just making things up randomly every day?" to my list of possible hypotheses.
I definitely agree there are probably some errors in the total predictions, I just want to know what is making the prediction change so dramatically!
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
To be honest, I’m not sure how much I trust their data. For example, in Michigan they are predicting peak deaths on April 9 (that part I’m not questioning, as it’s consistent with other reports I’ve seen). But then if you look at the graph, they are predicting no more deaths before the start of May. I just don’t see how it’s possible that 3 weeks from now, there will be no more deaths in MI. Even if we get things massively under control, and the hospital systems are not even close to overwhelmed at that point, I have to imagine there will still be deaths then. People are still being exposed today, and could be to the point of being in critical condition 3 weeks from now. Even with the best possible medical care, some percentage of cases will be fatal. And maybe I could be persuaded to think that just maybe deaths would be so few that they just don’t register on the scale of the graph, but their projected total deaths by August doesn’t show an increase at all, either. I would think the cumulative deaths by then would be showing a change in the line at some point after May 1. Even though I really wish we would be done with this by May, it makes me question the rest of their results.
Yes, they're updating every day and detailing how the models are changing with new information in their write-ups. Their noted margins of error around their curves are huge, so every possible piece of additional data helps to tighten things up. IHME is super well respected in the health analytics world. They do fascinating work.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
I don't trust that model either. It seems like some of their data is wrong. They're predicting a peak of 286 ventilators in Louisiana. According to our department of health there are 507 people on ventilators as of yesterday. The predicted peak for all beds is also less than the number of people currently hospitalized.
I don't trust that model either. It seems like some of their data is wrong. They're predicting a peak of 286 ventilators in Louisiana. According to our department of health there are 507 people on ventilators as of yesterday. The predicted peak for all beds is also less than the number of people currently hospitalized.
Same in Missouri. There are more people hospitalized and in the ICU in my metro than it says will be at peak for the whole state.
Yes, they're updating every day and detailing how the models are changing with new information in their write-ups. Their noted margins of error around their curves are huge, so every possible piece of additional data helps to tighten things up. IHME is super well respected in the health analytics world. They do fascinating work.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
A friend of mine who works for the local public hospital shared her analysis of what's going on with the numbers reported on that site. She said:
I've seen a couple people post about a specific model looking better (https://covid19.healthdata.org/projections…). I wanted to provide a bit of context on this and other models. 1) As our lead data guy working on our model says (quoting someone else) "all models are wrong, we just hope they will be useful" 2) The model here from IHME is a statistical model. This one uses the death rate to predict things moving forward. The model for Colorado has drastically changed within the last 7 days (from saying the peak would be this coming week or next week to saying it is now over. Statistical models aren't necessarily bad, but the inputs can drastically change the model. Colorado, for example, had an early spike in deaths due to "super spreader events" (e.g. nursing homes, bridge club, etc. outbreaks of the virus). This may have made our initial death rate much higher than expected. This model also assumes that social distancing kind of holds on constantly (which we know isn't true), so we would expect to have another peak in the Fall/Winter time as we allow more people to go back to work, etc. This model also bases off of California's population and assumes similar populations in other cities. 3) The model we have been doing at work is a mathematical model. It has lots of inputs including age-adjusted attack rates from the heavily referred to Imperial College paper (basically for each age group, who do we expect to recover at home, be hospitalized, need a ventilator, die, etc.). These could absolutely be wrong as could our assumptions on the R0 value (how many people one person is currently infecting after social distancing). Our model allows us to play with that time period to see when another spike could occur. Our model uses our local population (Denver, surrounding counties, Colorado, or whatever other state input you want to use). 4) Over the weekend, the White House identified Colorado as being a possible next national hot spot for the virus. So one model (which the White House was using) says we are past the peak while the White House tells us that we're an up and coming hot spot.
All this is to say that a model is wonderful and can help inform, and I am thrilled the death rate isn't rising as fast in Colorado. I am also seeing in our own hospital that the people who get on ventilators are on them a long time and many aren't showing up to a hospital until they are in desperate need (intubated in the emergency department), so I also expect the death rate (which is a lagging metric, for sure), may change a lot day to day as people are on ventilators for a long time trying to get them over the hump. Let's not assume we are past the worst of it. And let's continue to advocate for the testing, PPE, ventilators, medicine for allowing patients to be on ventilators, etc. that we need. We still don't want hospital staff to have to make the choice of who has the best chance of surviving.
This is fascinating from a scientific standpoint. A few days ago when I checked Colorado was going to be way under resourced (like <50% of needed beds) at our peak on April 15, and now it suggests we're already a day past our peak and will have enough beds.
I'm curious what is changing. Is social distancing working? Are we adding capacity? Is the model getting better?
I would trust this one more, since it's directly from our state DOH and the school of public health, which has more detailed data than what the healthdata.org page is pulling from.
I don't trust that model either. It seems like some of their data is wrong. They're predicting a peak of 286 ventilators in Louisiana. According to our department of health there are 507 people on ventilators as of yesterday. The predicted peak for all beds is also less than the number of people currently hospitalized.
::waves hi:: I’m in Louisiana also and I was coming in to say essentially the same thing...that I have trouble believing some of the newer, rosier projections. IDK. Even local media (I’m in NOLA) has articles saying numbers are starting to look better, but then saying...in literally the next sentence...that changes in methodology call the new projections into question. It’s like ‘okay so then...what do I do with that’?!?
One of the things I've noticed over the weeks is that weekends affect data numbers. So each monday it looks like maybe things are getting better or lowing down, or whatever,and then it looks like a new spike. trackers that average over time help with that.
Post by theoriginalbean on Apr 8, 2020 8:45:12 GMT -5
I really appreciate all of the modeling being done and the data collections and the analyses, but yesterday I started thinking that we're really never going to know the true magnitude of cases or deaths. There just isn't enough data being collected - there isn't enough testing, there isn't enough time. This article sums up the death count issue - www.cnn.com/2020/04/06/health/coronavirus-coroners-uncounted-deaths-invs/index.html - and really left me pessimistic that the count, while high and scary and incredibly tragic, is going to be looked at as "not that bad" and that America overreacted because it's going to be so artificially low.
I don't trust that model either. It seems like some of their data is wrong. They're predicting a peak of 286 ventilators in Louisiana. According to our department of health there are 507 people on ventilators as of yesterday. The predicted peak for all beds is also less than the number of people currently hospitalized.
::waves hi:: I’m in Louisiana also and I was coming in to say essentially the same thing...that I have trouble believing some of the newer, rosier projections. IDK. Even local media (I’m in NOLA) has articles saying numbers are starting to look better, but then saying...in literally the next sentence...that changes in methodology call the new projections into question. It’s like ‘okay so then...what do I do with that’?!?
Hi! BR here. We moved here from NOLA 2.5 years ago. Expected it to be temporary but it will be at least another 2 years now before we can get back down there. We miss it very much.
But yeah, it's hard to trust any of the data. Testing is not happening nearly enough, so I know the numbers are skewed. The major testing site in EBR parish is only testing 3 days a week, so that's going to cause spikes as those results come back, and days with less positives in between. Even if things are turning around in the NOLA area, I don't think we've hit a peak in BR yet. We just haven't tested nearly enough.
I would love to believe this projection that says there won't be a shortage of beds/equipment. I just don't believe it yet.
Overall consensus from people analyzing the data over time is that the model is actually overly pessimistic, not optimistic as many in this thread think. I think we all need to be very careful about consuming too much mainstream media because, like in the cases of many natural disasters, pessimism and doom and gloom sells. It keeps you tied to your television screen and scared.
I am not saying to ease up on social distancing. I actually think Dr. Fauci has found a good tone of being persistent in his plea to continue mitigation measures, but providing an optimistic outlook on the summer and fall. I also think it is appropriate to look at the death rate more than the case rate since testing is so low pretty much everywhere. The COVID19 board on reddit is only scientific papers so it helps me to read the science and not the latest vox or atlantic article.
I also think it is appropriate to look at the death rate more than the case rate since testing is so low pretty much everywhere.
Death rates are at least as under representative as positive case rates because under testing is even more of an issue.
If you don't die in the hospital being treated for COVID, or have already tested positive, then you aren't counted. Most places aren't taking samples from dead bodies. NYC has talked about this - many people have been found dead at home who are being collected and buried, but aren't in the coronavirus death counts.
Currently untested but still surviving cases might later be detected by antibody testing or might be tested tomorrow. Once you die that stops the clock for diagnosing the virus.
If you haven't already been tested and counted as positive before death, you never will be.
"The updated modeling results in this report continue to indicate that all control measures available, including relatively high levels of social distancing need to be utilized. Increases in case detection and isolation, mask-wearing and social distancing of approximately 65% can prevent a surge in infections in excess of hospital capacity in the coming summer months."
I have no idea how to calculate it, but even though you are opening does not mean you'll be at 100% (or 0 social distancing). There are no baseball games, fairs, concerts, etc. I'm assuming large weddings and birthday parties aren't happening. Capacity is limited in restaurants and stores so people can maintain distance.
I have no idea how to calculate it, but even though you are opening does not mean you'll be at 100% (or 0 social distancing). There are no baseball games, fairs, concerts, etc. I'm assuming large weddings and birthday parties aren't happening. Capacity is limited in restaurants and stores so people can maintain distance.
Time will tell shortly. Based on the behaviors I've seen, maintaining 65% social distancing is going to be hard. The cell phone data will confirm or not soon enough. Then we'll see if infection rates follow.
This was so interesting. Especially scrolling through on my phone b/c you really feel the scrolling distance between the 1.15x rates and like NY that is almost 6x.
Because it gives someone in their 40s with high blood pressure a worse survival chance than someone in their 80s with no preexisting conditions. It's freaking me out a bit about DH's high blood pressure :/