One of the increasing signs that the pandemic is on its way out here, other than the fewer people wearing masks outdoors, is that at the bookstore today the plastic screens at the counter were gone.
But of course they're still talking about cases, trying to make out that the virus, which they've finally decided to call Omicron, flew in to Nelson NZ from Auckland NZ. It's pure bullshit. They're not going to outrun the increasing desire to be had of this theater.
1. You've got spammers here - "jorden" and "Bailey" - they need banning, directly. Or at least that's what I do. I suppose if you get enough of it, you may have to go to a "subscribe to comment" mode...
2. I'd have your friend do similar analysis for other pandemics, like the 1918-1920 Spanish Flu, the 1968-1969 Hong Kong Flu, the 1975-1976 Swine Flu, the SARS and MERS epidemics, and see how well a curve can be fit.
3. It might be interesting to get in touch with Lubos Motl, a Czech theoretical physicist, who has done statistical analysis of this epidemic - a post of his here: https://motls.blogspot.com/2021/10/most-covid-deaths-are-not-due-to-covid.html And Motl knows about the Gompertz curve and its connection with epidemic modelling - "In the Parliament, Senator Dr Jan Žaloudík, an ex-director of the Masaryk Oncological Institute, gave a fiery speech chastising the Czech public TV that has "completely lost its mind" because it airs a 24-hour-per-day special broadcast about the Chinese flu virus even now, when the fairy-tale about a virus that threatens the whole planet, is collapsing like a house of cards. All of us have seen some curves of the growth when we were high school students. The Gompertz curve governs the sexual activity in one's life, ambitions of politicians, growth of rodents, Candida overgrowth, and also infections and epidemics." https://motls.blogspot.com/2020/04/pirk-covid-19-was-type-of-flu-for-me.html
Whilst the from/with distinction is important - and will lead to inflated figures there is going to be a relationship. I'm sure a lot of those deaths were hastened by covid - another insult to an already stressed system that tipped things over the edge.
Joel is looking at all-cause mortality to try to get around these sorts of issues.
However, what I find fascinating is that there is such a pattern to be found here at all. Riding on the top of "normal" death is this covid wave - and it's this that is generating the discernible pattern.
Yes, it is clear that there has been a change in allocation of cause of death. As I saw described only yesterday, CoD at end of natural span is full of subjectivity because of all the comorbities, proximal immediate causes v causal disorders etc. But when we fly higher up we can rise above that variability and see that there is no difference in the totals and the seasonal wave shape over time. Sanity, sweet sanity. Same balloon, squeezed in different ways on death certs. Used to be the flux, rising of the lights, dropsy.
This change in allocation is quite obvious when you look pneumonia before and after the adoption of ICD-10 (2001), and an accompanying new selection rule for underlying cause of death. Around half of deaths previously attributed to pneumonia moved to other causes.
It isn't a problem. There is no need to disentangle anything. It's impossible anyway as the data is debauched. The point here is to look at what signal there is in early 20 and 21 and 22 that distinguishes it from all cause mortality in previous years. If it's simply a reallocation of cause of death within the same numeric envelope then...just what is all the fuss about.
I agree that the Gompertz models are exciting. But they are models nevertheless. And as Briggs never tires of pointing out, models have three characteristics: (1) they are beautiful to look at, (2) they are completely artifical, and (3) they only do what they are told to do. Models have to be judged by their usefulness. Let the models make predictions, and then compare to what happens.
Oh, and in the "galley images" I miss oars and someone banging the drum. Or, as David Lodge wrote about a visit to the "Tropical Waterworld" in "Deaf Sentence": "Change the soundtrack, substitute screams and howls for laughter and badinage, put a red filter on the lens to give a fiery glow to the spectacle, and you would think you were in some modern version of Dante’s Inferno, or the hells depicted by medieval painters."
Yes, models should always be used sensibly. Joel's modelling here is an empirical "data fit" model - it might be possible to make predictions from it, but I am not sure how that could be done at present. Joel might have some ideas on that.
The fact that there exists such a simple model that can be beautifully fit to the data - and with a highly suggestive temporal relationship (highly suggestive, but not proof of, causality) - is intriguing to me. It might, of course, be entirely accidental that such an underlying pattern exists, but I don't think so.
One of the increasing signs that the pandemic is on its way out here, other than the fewer people wearing masks outdoors, is that at the bookstore today the plastic screens at the counter were gone.
But of course they're still talking about cases, trying to make out that the virus, which they've finally decided to call Omicron, flew in to Nelson NZ from Auckland NZ. It's pure bullshit. They're not going to outrun the increasing desire to be had of this theater.
1. You've got spammers here - "jorden" and "Bailey" - they need banning, directly. Or at least that's what I do. I suppose if you get enough of it, you may have to go to a "subscribe to comment" mode...
2. I'd have your friend do similar analysis for other pandemics, like the 1918-1920 Spanish Flu, the 1968-1969 Hong Kong Flu, the 1975-1976 Swine Flu, the SARS and MERS epidemics, and see how well a curve can be fit.
3. It might be interesting to get in touch with Lubos Motl, a Czech theoretical physicist, who has done statistical analysis of this epidemic - a post of his here: https://motls.blogspot.com/2021/10/most-covid-deaths-are-not-due-to-covid.html And Motl knows about the Gompertz curve and its connection with epidemic modelling - "In the Parliament, Senator Dr Jan Žaloudík, an ex-director of the Masaryk Oncological Institute, gave a fiery speech chastising the Czech public TV that has "completely lost its mind" because it airs a 24-hour-per-day special broadcast about the Chinese flu virus even now, when the fairy-tale about a virus that threatens the whole planet, is collapsing like a house of cards. All of us have seen some curves of the growth when we were high school students. The Gompertz curve governs the sexual activity in one's life, ambitions of politicians, growth of rodents, Candida overgrowth, and also infections and epidemics." https://motls.blogspot.com/2020/04/pirk-covid-19-was-type-of-flu-for-me.html
4. Here's a review article which cites in references a number of predictive models - although the article doesn't explicitly mention the Gompertz curve - https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC5159251/ - and here's an article in Spanish which uses Gompertz' analysis - https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC7256556/
so joel figured out how to separate the deaths from covid from the deaths with covid?
for if he didn't , all he did was make a mathematical model of something rather irrelevant...
Whilst the from/with distinction is important - and will lead to inflated figures there is going to be a relationship. I'm sure a lot of those deaths were hastened by covid - another insult to an already stressed system that tipped things over the edge.
Joel is looking at all-cause mortality to try to get around these sorts of issues.
However, what I find fascinating is that there is such a pattern to be found here at all. Riding on the top of "normal" death is this covid wave - and it's this that is generating the discernible pattern.
Yes, it is clear that there has been a change in allocation of cause of death. As I saw described only yesterday, CoD at end of natural span is full of subjectivity because of all the comorbities, proximal immediate causes v causal disorders etc. But when we fly higher up we can rise above that variability and see that there is no difference in the totals and the seasonal wave shape over time. Sanity, sweet sanity. Same balloon, squeezed in different ways on death certs. Used to be the flux, rising of the lights, dropsy.
This change in allocation is quite obvious when you look pneumonia before and after the adoption of ICD-10 (2001), and an accompanying new selection rule for underlying cause of death. Around half of deaths previously attributed to pneumonia moved to other causes.
problem is that the relationship varies widely over time and place.
unless one can establish what part of the total deaths is " normal " and what part is
" covid wave " it is impossible to reach any conclusions.
It isn't a problem. There is no need to disentangle anything. It's impossible anyway as the data is debauched. The point here is to look at what signal there is in early 20 and 21 and 22 that distinguishes it from all cause mortality in previous years. If it's simply a reallocation of cause of death within the same numeric envelope then...just what is all the fuss about.
I agree that the Gompertz models are exciting. But they are models nevertheless. And as Briggs never tires of pointing out, models have three characteristics: (1) they are beautiful to look at, (2) they are completely artifical, and (3) they only do what they are told to do. Models have to be judged by their usefulness. Let the models make predictions, and then compare to what happens.
Oh, and in the "galley images" I miss oars and someone banging the drum. Or, as David Lodge wrote about a visit to the "Tropical Waterworld" in "Deaf Sentence": "Change the soundtrack, substitute screams and howls for laughter and badinage, put a red filter on the lens to give a fiery glow to the spectacle, and you would think you were in some modern version of Dante’s Inferno, or the hells depicted by medieval painters."
Yes, models should always be used sensibly. Joel's modelling here is an empirical "data fit" model - it might be possible to make predictions from it, but I am not sure how that could be done at present. Joel might have some ideas on that.
The fact that there exists such a simple model that can be beautifully fit to the data - and with a highly suggestive temporal relationship (highly suggestive, but not proof of, causality) - is intriguing to me. It might, of course, be entirely accidental that such an underlying pattern exists, but I don't think so.
Michael Levitt seems to work at this all the time. I will save this image and see how things turn out: https://twitter.com/MLevitt_NP2013/status/1485028245229776910/photo/1