Coronavirus - Modelling Aspects Only

[Mike4]

and their vitamin D levels were good despite their skin colour or because they take supplements?

I give advice on supplements including D3. The levels were good because they follow my advice. I generally recommend 3000iu a day, but I think that can be too much in the long term if people are white and not obese.

Hence what I am saying is that the doctors though the outcome from my advice was good.

I just follow advice from the NHS (as given by my Sister-in-law GP):

https://www.nhs.uk/conditions/vitamins- ... vitamin-d/

I.e. 10mg (400IU) per day.

400iu per day is widely considered a woefully inadequate supplement by doctors. It is a dose based on the D levels required for good bones (AIUI). One doctor I heard talking about the NHS advised dose said 400iu "won't even touch the sides" if given to someone deficient in D and needing it for restoring their immune function.

This dose was arrived at before the importance of D for immune function was fully understood and has never been changed. Again AIUI. 3000iu per day is the dose I repeatedly hear doctors say they take themselves, but always qualified with the comment they cannot give medical advice over the internet so that dose may not be appropriate for you.

[johnhemming]

400iu per day is widely considered a woefully inadequate supplement by doctors. It is a dose based on the D levels required for good bones (AIUI). One doctor I heard talking about the NHS advised dose said 400iu "won't even touch the sides" if given to someone deficient in D and needing it for restoring their immune function.

This dose was arrived at before the importance of D for immune function was fully understood and has never been changed. Again AIUI. 3000iu per day is the dose I repeatedly hear doctors say they take themselves, but always qualified with the comment they cannot give medical advice over the internet so that dose may not be appropriate for you.

Germany recommends 800iu.

I am experimenting with various levels and recording outcomes (particularly for sleep). Today, for example, I have decided to stop taking 3000iu and to take none for a few days whilst my levels go down. I think for my current weight 3000 is probably too high. I have, however, lost quite a bit of weight recently and vitamin D can be dissolved in fat so the more fat you have the more vitamin D you need (it appears).

Hence if you are in the normal weight range I think 3000 as a maintenance dose is probably too high for people who have a white skin. However, it is the sort of thing that has an elimination half life of around 42 days (1000 hours) so taking a loading dose is a reasonable thing to do, but then after a while people may need to reduce their dose. That will vary by individual.

[funduffer]

400iu per day is widely considered a woefully inadequate supplement by doctors. It is a dose based on the D levels required for good bones (AIUI). One doctor I heard talking about the NHS advised dose said 400iu "won't even touch the sides" if given to someone deficient in D and needing it for restoring their immune function.

This dose was arrived at before the importance of D for immune function was fully understood and has never been changed. Again AIUI. 3000iu per day is the dose I repeatedly hear doctors say they take themselves, but always qualified with the comment they cannot give medical advice over the internet so that dose may not be appropriate for you.

Germany recommends 800iu.

I am experimenting with various levels and recording outcomes (particularly for sleep). Today, for example, I have decided to stop taking 3000iu and to take none for a few days whilst my levels go down. I think for my current weight 3000 is probably too high. I have, however, lost quite a bit of weight recently and vitamin D can be dissolved in fat so the more fat you have the more vitamin D you need (it appears).

Hence if you are in the normal weight range I think 3000 as a maintenance dose is probably too high for people who have a white skin. However, it is the sort of thing that has an elimination half life of around 42 days (1000 hours) so taking a loading dose is a reasonable thing to do, but then after a while people may need to reduce their dose. That will vary by individual.

Thanks for the interesting discussion on this.

I suspect the optimum dose very much depends on the individual - the colour of their skin, how much they go outside during daylight hours, how much of their skin is exposed, their fat level and maybe a few other factors. It is the first time I have heard that the NHS recommendation is very conservative.

Luckily I have a fairly good 'Vitamin D receptor' on my head (I am bald) and get out in the daylight every day, so maybe a lowish dose is right for me?

FD

[johnhemming]

Luckily I have a fairly good 'Vitamin D receptor' on my head (I am bald) and get out in the daylight every day, so maybe a lowish dose is right for me?

I would think that is unlikely in the UK.

https://www.nhs.uk/live-well/healthy-bo ... -sunlight/

Winter sunlight
In the UK, sunlight doesn't contain enough UVB radiation in winter (October to early March) for our skin to be able to make vitamin D.

During these months, we rely on getting our vitamin D from food sources (including fortified foods) and supplements.

Using sunbeds isn't a recommended way of making vitamin D.

[scotia]

I have been keeping a watch at the Covid-19 death rates in England, and have been pleased to see that growth rate has been significantly moderating.

To recap - each data point is the sum of the preceding week of data, normalised to a starting point of 1. I have plotted the Natural Log of this ratio and for any straight line segment, the gradient can be directly read as the exponential growth factor in days (Lambda). The associated doubling time is ln(2)/Lambda. The vertical lengths of the lines on the data points are plus and minus the standard deviations, assuming a Poisson distribution.
In my previous graphs with data to the week ending 25/10/19 the doubling time was 10.8 days and with data to the week ending of 1/11/20 it was 12.46. In the above graph - to the week ending 1/11/20 it is 19.08. Good News!

And If I now overlay real data on my earlier graphical projections, I now see that I was unduly pessimistic.


So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

[johnhemming]

So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

This I think is a good data source
https://app.powerbi.com/view?r=eyJrIjoi ... IsImMiOjh9

If you click "excluded" to exclude those which clearly don't have Covid you can see that GP consultations peaked in the week recorded as 26/10.

Given that deaths will generally follow consultations by around 3 weeks we are getting quite a bit of information pointing to the same thing. Hospital admissions

https://coronavirus.data.gov.uk/details ... me=England

Have gone up a bit recently, but again seem to fit in with the same pattern (given delays between sets of data)

[scotia]

Now if you really want to see an impressive site displaying Covid-19 statistics , try
https://www.travellingtabby.com/scotland-coronavirus-tracker/
It is, however, only for Scotland. Maybe you would like to offer the student, who produced it, a bribe to include England.

[zico]

I have been keeping a watch at the Covid-19 death rates in England, and have been pleased to see that growth rate has been significantly moderating.

So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

Would be great if the number of daily deaths reduced, but what's the underlying logic behind your model predictions? The R-value is still above 1, so the simplest model is to assume that hospitalisations and deaths will increase in future weeks if the number of infections is increasing.

[johnhemming]

This comes to the question as to whether the government have reliable numbers.

Some numbers are more reliable than others. I don't think their R figure means much at all.

[Nimrod103]

I have been keeping a watch at the Covid-19 death rates in England, and have been pleased to see that growth rate has been significantly moderating.

So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

Would be great if the number of daily deaths reduced, but what's the underlying logic behind your model predictions? The R-value is still above 1, so the simplest model is to assume that hospitalisations and deaths will increase in future weeks if the number of infections is increasing.

From Dr John Campbell's Youtube video today - SAGE maintain that the UK's R value is still 1.1-1.2, but the Covid symptom tracker app of Kings College (which has proved very accurate to date) says 0.9, and infections are now on the decline. Deaths will probably peak in a fortnight.

[Lootman]

I have been keeping a watch at the Covid-19 death rates in England, and have been pleased to see that growth rate has been significantly moderating.

So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

Would be great if the number of daily deaths reduced, but what's the underlying logic behind your model predictions? The R-value is still above 1, so the simplest model is to assume that hospitalisations and deaths will increase in future weeks if the number of infections is increasing.

From Dr John Campbell's Youtube video today - SAGE maintain that the UK's R value is still 1.1-1.2, but the Covid symptom tracker app of Kings College (which has proved very accurate to date) says 0.9, and infections are now on the decline. Deaths will probably peak in a fortnight.

Didn't deaths peak last April/May timeframe?

I thought that is the real point here, that cases are higher but deaths are lower, indicating one or more of a few things:

1) Treatments are getting better, and/or

2) The weakest and most susceptible people have already been killed off, and/or

3) The virus is weakening for reasons we do not yet understand, and/or

4) People are building up resistance and/or immunity

[Mike4]

I thought that is the real point here, that cases are higher but deaths are lower, indicating one or more of a few things:

1) Treatments are getting better, and/or

2) The weakest and most susceptible people have already been killed off, and/or

3) The virus is weakening for reasons we do not yet understand, and/or

4) People are building up resistance and/or immunity

5) We are detecting more of the infections that were/are happening anyway.


Also depends whether by "case". Do you mean the layman's definition, i.e. a positive PCR test result or what the doctors mean, i.e. someone with symptoms of COVID-19.

[swill453]

Also depends whether by "case". Do you mean the layman's definition, i.e. a positive PCR test result or what the doctors mean, i.e. someone with symptoms of COVID-19.

Most people with Covid-19 won't see a doctor.

Scott.

[Mike4]

Also depends whether by "case". Do you mean the layman's definition, i.e. a positive PCR test result or what the doctors mean, i.e. someone with symptoms of COVID-19.

Most people with Covid-19 won't see a doctor.

Scott.

Most people infected asymptomatically won't see a doctor either.

[swill453]

Also depends whether by "case". Do you mean the layman's definition, i.e. a positive PCR test result or what the doctors mean, i.e. someone with symptoms of COVID-19.

Most people with Covid-19 won't see a doctor.

Most people infected asymptomatically won't see a doctor either.

They're included in my reply.

Scott.

[scotia]

I have been keeping a watch at the Covid-19 death rates in England, and have been pleased to see that growth rate has been significantly moderating.

So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

Would be great if the number of daily deaths reduced, but what's the underlying logic behind your model predictions? The R-value is still above 1, so the simplest model is to assume that hospitalisations and deaths will increase in future weeks if the number of infections is increasing.

The logic behind the model is the assumption that the lockdown produces a similar fall in death rate as in April. I estimated that to be around a 14 day halving period - but I also plotted 11 and 17 day halving periods. That may be optimistic - its not a full lockdown, and maybe the populace is no longer as willing to obey the regulations as they were in April.

[GoSeigen]

So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

Says the poster who was predicting 10000 deaths per week by the end of November, then got my posts kicked off this board when I had the temerity to dispute the output of his "model".

Incidentally even 10,000 per week would be barely above the April peak so even if it had occurred that would hardly qualify as a new wave of the pandemic. The danger point of a pandemic is when it is spreading exponentially: it was clear by July that that phase was over. The remaining deaths, including what some have misnamed a "second wave" constitute a slow arithmetic growth rate, which is not scary and quite manageable.

GS

[johnhemming]

The remaining deaths, including what some have misnamed a "second wave" constitute a slow arithmetic growth rate, which is not scary and quite manageable.

I think it is fair to say that initially it was exponential then moved to linear and then peaked.

That is what one would expect really for a situation where a seaonably variable virus moves into a period of greater virulence. The good news (which was uncertain) is that it appears now to have hit a form of equilibrium. To what extent that is across the northern hemisphere we will know with some certainty in a couple of weeks.

[scotia]

So I would like to be optimistic, and think we should see the death rate peaking by the end of the coming week.

Says the poster who was predicting 10000 deaths per week by the end of November, then got my posts kicked off this board when I had the temerity to dispute the output of his "model".

Incidentally even 10,000 per week would be barely above the April peak so even if it had occurred that would hardly qualify as a new wave of the pandemic. The danger point of a pandemic is when it is spreading exponentially: it was clear by July that that phase was over. The remaining deaths, including what some have misnamed a "second wave" constitute a slow arithmetic growth rate, which is not scary and quite manageable.

GS

Why do you insist on publishing false claims? So once again, I will repeat exactly what I said, after plotting the Covid-19 Deaths data, and fitting them to an exponential on October 19th
"If the data continues to follow this exponential curve, then in a further 40 days there will be around 10,000 deaths per week"
That statement was and is mathematically accurate. If I remember correctly your "mathematics" (in a science and modelling board) claimed that the number was 2000 - and I asked, but got no response as to how your "mathematics" obtained that number from the plotted exponential curve.
Fortunately government actions were taken, particularly in the North of England, to reduce the infection reproduction rate, and hence the exponential growth rate. I continued to plot the data with fits to this improving rate, and reported the results.

Now lets return to your mathematics. Would you care to explain what your "arithmetic growth rate" means? Would you like provide your equation for such feature, and would you care to plot a curve and fit it to the observed data? I have asked for such a response before, but have so far had no response. Maybe today?

And finally, I did not get your posts booted off this board. I did not flag your posts to a moderator. However since your posts were clearly not science nor were they models, I'm not surprised that your posts were removed.

[redsturgeon]

Moderator Message:
I have just moved the last two posts to the Snug thread. They were off topic, not on modelling.