[BostonCard]

Nice tool to explore the SEIR model, which is the backbone of infectious disease models.

http://gabgoh.github.io/COVID/index.html

Remember, George Box' aphorism that "all models are wrong; some models are useful."

BC

[BostonCard]

Nice tool to explore the SEIR model, which is the backbone of infectious disease models.

http://gabgoh.github.io/COVID/index.html

Remember, George Box' aphorism that "all models are wrong; some models are useful."

BC

[martyup]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

[martyup]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

[BostonCard]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

I don't know; it almost certainly depends on how well people are following the social distancing. Intuitively, I would like to think it is less than 1, hopefully significantly so, but I really don't know.

In a few weeks, once we've had a chance to see how the experiment works, it will be possible to re-fit the model and recalculate the R(eff) with the social distancing measures in place.

BC

[BostonCard]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

I don't know; it almost certainly depends on how well people are following the social distancing. Intuitively, I would like to think it is less than 1, hopefully significantly so, but I really don't know.

In a few weeks, once we've had a chance to see how the experiment works, it will be possible to re-fit the model and recalculate the R(eff) with the social distancing measures in place.

BC

[M T]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

I don't know; it almost certainly depends on how well people are following the social distancing. Intuitively, I would like to think it is less than 1, hopefully significantly so, but I really don't know.

In a few weeks, once we've had a chance to see how the experiment works, it will be possible to re-fit the model and recalculate the R(eff) with the social distancing measures in place.

BC

Wouldn't the data of some of the other countries gives us at least a lower bound, as our SiP order isn't as strict as their quarantine?
Some super spreaders would be grocery clerks, delivery personnel, ...
There are probably pockets of people not keeping their distance (homeless, illegal drug users, etc) that may still be spreading the disease.
And, of course, there will be spread within a household.

Last I looked this afternoon, there were no Santa Clara County numbers up for yesterday.  (I'm sure they're plenty busy).  I'm eagerly awaiting as I hope the
number of new cases should start dropping soon.

[M T]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

I don't know; it almost certainly depends on how well people are following the social distancing. Intuitively, I would like to think it is less than 1, hopefully significantly so, but I really don't know.

In a few weeks, once we've had a chance to see how the experiment works, it will be possible to re-fit the model and recalculate the R(eff) with the social distancing measures in place.

BC

Wouldn't the data of some of the other countries gives us at least a lower bound, as our SiP order isn't as strict as their quarantine?
Some super spreaders would be grocery clerks, delivery personnel, ...
There are probably pockets of people not keeping their distance (homeless, illegal drug users, etc) that may still be spreading the disease.
And, of course, there will be spread within a household.

Last I looked this afternoon, there were no Santa Clara County numbers up for yesterday.  (I'm sure they're plenty busy).  I'm eagerly awaiting as I hope the
number of new cases should start dropping soon.

[BostonCard]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

I don't know; it almost certainly depends on how well people are following the social distancing. Intuitively, I would like to think it is less than 1, hopefully significantly so, but I really don't know.

In a few weeks, once we've had a chance to see how the experiment works, it will be possible to re-fit the model and recalculate the R(eff) with the social distancing measures in place.

BC

Wouldn't the data of some of the other countries gives us at least a lower bound, as our SiP order isn't as strict as their quarantine?
Some super spreaders would be grocery clerks, delivery personnel, ...
There are probably pockets of people not keeping their distance (homeless, illegal drug users, etc) that may still be spreading the disease.
And, of course, there will be spread within a household.

Last I looked this afternoon, there were no Santa Clara County numbers up for yesterday.  (I'm sure they're plenty busy).  I'm eagerly awaiting as I hope the
number of new cases should start dropping soon.

As I noted at the onset, all models are wrong, some models are useful.  One drawback of the SEIR model is that it assumes that all contacts are of equal probability.  That is, in SEIR you are just as likely to come into contact with a stranger across the country as with your spouse. We obviously know that's wrong, and in reality you have circles of decreasing contact (household contacts are very strong, followed by work/school contacts, followed by random contacts locally, followed by even broader contacts). 

SEIR would assume that the epidemic was occurring homogeneously across the US; in truth there are hotspots and some places where it is less common.  But it turns out that just modeling average contact works pretty well.

BC

[BostonCard]

This is amazing.  What is the most likely number for Rt with the social distancing approach being taken by most states?

I don't know; it almost certainly depends on how well people are following the social distancing. Intuitively, I would like to think it is less than 1, hopefully significantly so, but I really don't know.

In a few weeks, once we've had a chance to see how the experiment works, it will be possible to re-fit the model and recalculate the R(eff) with the social distancing measures in place.

BC

Wouldn't the data of some of the other countries gives us at least a lower bound, as our SiP order isn't as strict as their quarantine?
Some super spreaders would be grocery clerks, delivery personnel, ...
There are probably pockets of people not keeping their distance (homeless, illegal drug users, etc) that may still be spreading the disease.
And, of course, there will be spread within a household.

Last I looked this afternoon, there were no Santa Clara County numbers up for yesterday.  (I'm sure they're plenty busy).  I'm eagerly awaiting as I hope the
number of new cases should start dropping soon.

As I noted at the onset, all models are wrong, some models are useful.  One drawback of the SEIR model is that it assumes that all contacts are of equal probability.  That is, in SEIR you are just as likely to come into contact with a stranger across the country as with your spouse. We obviously know that's wrong, and in reality you have circles of decreasing contact (household contacts are very strong, followed by work/school contacts, followed by random contacts locally, followed by even broader contacts). 

SEIR would assume that the epidemic was occurring homogeneously across the US; in truth there are hotspots and some places where it is less common.  But it turns out that just modeling average contact works pretty well.

BC

[M T]

A friend pointed out a couple of papers

"Fundamental principles of epidemic spread highlight the immediate need for large-scale serological surveys to assess the stage of the SARS-CoV-2 epidemic"  from Oxford.  Mar 26.

WSJ commentary:(I see multiple copies reposted around the net without the pay wall.)
 "Is the Coronavirus as Deadly as They Say?" by Eran Bendavid and Jay Bhattacharya of Stanford.  Mar 24.

The first considers that there may have been thousands of very mild/asymptomatic cases in the UK before the first death.  The second asks whether there are millions of undetected cases, which then drives the CFR to a low number.

My issues with both these papers are
  1)  AFAICT, they don't take into account the high number of negative test results, for instance, the 320k tests in Guangdong province with just 0.47% positive results.  If the disease was spreading widely under the radar, don't you think they would have stumbled onto clusters of the disease.
   2) Apparently if you test positive with the current test, there is a 20% chance of pneumonia, and an 80% chance of no pneumonia.  This isn't Schrödinger's cat.  If you get the disease that this test is detecting, then you have a decent chance of getting pneumonia.   If there were thousands of cases in the UK, then that would be hundreds of pneumonia cases that have a particular characteristic.  It is hard for me to believe that there would not have been some cluster that would have been spotted.   But it wouldn't have been just the UK, but China, the US,  S. Korea, etc.  No one spotted it?

A scenario I could imagine is a cow pox/smallpox analogy.  There could have been some disease that swept through much of the world that no one paid much attention to because it was mild.  However, the antibodies it produced work to block or minimize COVID-19.   Many people could in theory have these antibodies.  The ones that didn't get the first disease are the 20% that are getting hit so hard, or maybe the 100% that are showing the symptoms.     BUT, the problem with this is that we aren't seeing COVID-19 impact the very young that surely didn't get the milder disease.

While I don't currently agree with these papers, they are from reputable places, so I wanted to bring them up here, for others to analyze, and maybe show me why I'm wrong.

[M T]

A friend pointed out a couple of papers

"Fundamental principles of epidemic spread highlight the immediate need for large-scale serological surveys to assess the stage of the SARS-CoV-2 epidemic"  from Oxford.  Mar 26.

WSJ commentary:(I see multiple copies reposted around the net without the pay wall.)
 "Is the Coronavirus as Deadly as They Say?" by Eran Bendavid and Jay Bhattacharya of Stanford.  Mar 24.

The first considers that there may have been thousands of very mild/asymptomatic cases in the UK before the first death.  The second asks whether there are millions of undetected cases, which then drives the CFR to a low number.

My issues with both these papers are
  1)  AFAICT, they don't take into account the high number of negative test results, for instance, the 320k tests in Guangdong province with just 0.47% positive results.  If the disease was spreading widely under the radar, don't you think they would have stumbled onto clusters of the disease.
   2) Apparently if you test positive with the current test, there is a 20% chance of pneumonia, and an 80% chance of no pneumonia.  This isn't Schrödinger's cat.  If you get the disease that this test is detecting, then you have a decent chance of getting pneumonia.   If there were thousands of cases in the UK, then that would be hundreds of pneumonia cases that have a particular characteristic.  It is hard for me to believe that there would not have been some cluster that would have been spotted.   But it wouldn't have been just the UK, but China, the US,  S. Korea, etc.  No one spotted it?

A scenario I could imagine is a cow pox/smallpox analogy.  There could have been some disease that swept through much of the world that no one paid much attention to because it was mild.  However, the antibodies it produced work to block or minimize COVID-19.   Many people could in theory have these antibodies.  The ones that didn't get the first disease are the 20% that are getting hit so hard, or maybe the 100% that are showing the symptoms.     BUT, the problem with this is that we aren't seeing COVID-19 impact the very young that surely didn't get the milder disease.

While I don't currently agree with these papers, they are from reputable places, so I wanted to bring them up here, for others to analyze, and maybe show me why I'm wrong.