4 ms·

What is the source for the doublerate ? Related to R0 ?

by bsg75 7y ago

dredmorbius 7y ago
From the Wikipedia Wuhan 2019-nCoV article, semi-log graph. Caption: Semi-log plot of confirmed cases and deaths indicates the epidemic is in an exponential phase. Doubling time is 1.71 days (95% confidence interval 1.64 to 1.79) for case numbers https://en.m.wikipedia.org/wiki/2019–20_Wuhan_coronavirus_outbreak#/media/File%3ALog-linear_plot_of_coronovirus_cases_with_linear_regressions.png https://en.m.wikipedia.org/wiki/2019–20_Wuhan_coronavirus_ou... Since there are low and high estimates, I could include those values as well in my model. OK, done. Note that this really shows how sensitive growth rates are to even modest changes in doubling rate. Note that the 'lo' column refers to the time to doubling, not the rate of doubling, which may otherwise be counterintuitive. In the 100-day run, the difference between a 1.64 day doubling rate vs. a 1.79 day doubling rate is total global infection by April 8, vs. still < 360 million infected by May 6. The real lesson here is that growth rate matters vastly more than starting quantity. That's something any investor should recognise innately, but it applies accross numerous other fields. Including pandemic spread, population growth, and resource utilisation rates of increase. | day | date | deaths | cases | @ lo dbl | @ hi dbl | |:--- | ------------:| ------------:| -------------:| -------------:| -------------:| | 1 | Jan 29, 2020 | 154 | 7,091 | 7,393 | 6,774 | | 2 | Jan 30, 2020 | 180 | 8,293 | 9,016 | 7,569 | | 3 | Jan 31, 2020 | 211 | 9,700 | 10,996 | 8,457 | | 4 | Feb 01, 2020 | 247 | 11,345 | 13,410 | 9,449 | | 5 | Feb 02, 2020 | 288 | 13,269 | 16,353 | 10,557 | | 6 | Feb 03, 2020 | 337 | 15,519 | 19,943 | 11,796 | | 7 | Feb 04, 2020 | 395 | 18,152 | 24,321 | 13,180 | | 8 | Feb 05, 2020 | 462 | 21,230 | 29,660 | 14,726 | | 9 | Feb 06, 2020 | 540 | 24,830 | 36,171 | 16,454 | | 10 | Feb 07, 2020 | 632 | 29,041 | 44,111 | 18,384 | | 11 | Feb 08, 2020 | 739 | 33,967 | 53,794 | 20,541 | | 12 | Feb 09, 2020 | 864 | 39,727 | 65,602 | 22,951 | | 13 | Feb 10, 2020 | 1,011 | 46,465 | 80,003 | 25,644 | | 14 | Feb 11, 2020 | 1,183 | 54,345 | 97,564 | 28,652 | | day | date | deaths | cases | @ lo dbl | @ hi dbl | |:--- | ------------:| ------------:| -------------:| -------------:| -------------:| | 1 | Jan 29, 2020 | 154 | 7,091 | 7,393 | 6,774 | | 8 | Feb 05, 2020 | 462 | 21,230 | 29,660 | 14,726 | | 15 | Feb 12, 2020 | 1,383 | 63,561 | 118,981 | 32,014 | | 22 | Feb 19, 2020 | 4,143 | 190,297 | 477,290 | 69,595 | | 29 | Feb 26, 2020 | 12,403 | 569,731 | 1,914,634 | 151,293 | | 36 | Mar 04, 2020 | 37,135 | 1,705,718 | 7,680,489 | 328,895 | | 43 | Mar 11, 2020 | 111,181 | 5,106,750 | 30,810,015 | 714,985 | | 50 | Mar 18, 2020 | 332,865 | 15,289,096 | 123,593,305 | 1,554,303 | | 57 | Mar 25, 2020 | 996,564 | 45,774,017 | 495,790,250 | 3,378,893 | | 64 | Apr 01, 2020 | 2,983,613 | 137,042,800 | 1,988,845,356 | 7,345,363 | | 71 | Apr 08, 2020 | 8,932,639 | 410,292,352 | 7,978,184,014 | 15,968,056 | | 78 | Apr 15, 2020 | 26,743,422 | 1,228,374,011 | 32,004,207,853 | 34,712,893 | | 85 | Apr 22, 2020 | 80,067,114 | 3,677,628,161 | 128,383,767,336 | 75,462,218 | | 92 | Apr 29, 2020 | 239,712,883 | 11,010,448,584 | 515,007,020,041 | 164,047,010 | | 99 | May 06, 2020 | 717,676,247 | 32,964,174,931 | 2,065,932,759,210 | 356,621,130 | Chart is formatted as a Markdown chart, BTW.
- perl4ever 7y ago
  "growth rate matters vastly more" Only in the case where you deny the existence of uncertainty in that number. Also, it's generally considered that infectious and lethal diseases rapidly evolve to become less virulent, since the people that are sickest tend to be more isolated.
  - dredmorbius 7y ago
    No. The point is that initial quantity has virtualy no impact on final result, as compared with growth rate. With a doubling time of under two days, a misestimation by a factor of two ... is resolved in a day. By a factor of 10, in about 3 days. By a factor of 100, in six. By a factor of 1,000, in 12. It's not uncertainty about the growth rate, but change which matters, and which epidemiological containment is concerned with. The goal is to change the transmission rate (R factor), and through that, the rate of growth of the epidemic. Epidemics of the past which have proved highly lethal (look through your favourite list of historical plagues) have killed from 10% to half of local populations, often through a combination of high mortality but slow onset. Spreading before* symptoms are clearly expressed will suffice. The Black Death spread throughout Europe from 1346 - 1353, over eight years, killing up to 60% of the population. It returned multiple times, through 1667, with recurrence mortalities of 10-20%. It continued through the 19th century in the Arabic world. Expecting a novel virus to attenuate markedly in a few days or weeks seems wishful thinking.
    - perl4ever 7y ago
      "It's not uncertainty about the growth rate, but change which matters" You're saying growth rate is not rate of change and the rate of change is important, but the difference between two different rates of change is insignificant? My point was that the R factor is neither constant nor precisely known, whatever people do. If the outcome is very sensitive to the number, then it seems to me obvious that it is very important to look at the outcome over the range of possibilities, because it's going to be a large range. Whenever professionals make long term predictions, they usually depict them as a cone. How big it is and how it widens is important. Also, I thought the plague was bacterial, so I'm doubtful of the inference that a virus won't evolve rapidly. Previous flu epidemics did, right?

What is the source for the doublerate ? Related to R0 ?

by bsg75 7y ago

What is the source for the doublerate ? Related to R0 ?

From the Wikipedia Wuhan 2019-nCoV article, semi-log graph. Caption: Semi-log plot of confirmed cases and deaths indicates the epidemic is in an exponential phase. Doubling time is 1.71 days (95% confidence interval 1.64 to 1.79) for case numbers https://en.m.wikipedia.org/wiki/2019–20_Wuhan_coronavirus_outbreak#/media/File%3ALog-linear_plot_of_coronovirus_cases_with_linear_regressions.png https://en.m.wikipedia.org/wiki/2019–20_Wuhan_coronavirus_ou... Since there are low and high estimates, I could include those values as well in my model. OK, done. Note that this really shows how sensitive growth rates are to even modest changes in doubling rate. Note that the 'lo' column refers to the time to doubling, not the rate of doubling, which may otherwise be counterintuitive. In the 100-day run, the difference between a 1.64 day doubling rate vs. a 1.79 day doubling rate is total global infection by April 8, vs. still < 360 million infected by May 6. The real lesson here is that growth rate matters vastly more than starting quantity. That's something any investor should recognise innately, but it applies accross numerous other fields. Including pandemic spread, population growth, and resource utilisation rates of increase. | day | date | deaths | cases | @ lo dbl | @ hi dbl | |:--- | ------------:| ------------:| -------------:| -------------:| -------------:| | 1 | Jan 29, 2020 | 154 | 7,091 | 7,393 | 6,774 | | 2 | Jan 30, 2020 | 180 | 8,293 | 9,016 | 7,569 | | 3 | Jan 31, 2020 | 211 | 9,700 | 10,996 | 8,457 | | 4 | Feb 01, 2020 | 247 | 11,345 | 13,410 | 9,449 | | 5 | Feb 02, 2020 | 288 | 13,269 | 16,353 | 10,557 | | 6 | Feb 03, 2020 | 337 | 15,519 | 19,943 | 11,796 | | 7 | Feb 04, 2020 | 395 | 18,152 | 24,321 | 13,180 | | 8 | Feb 05, 2020 | 462 | 21,230 | 29,660 | 14,726 | | 9 | Feb 06, 2020 | 540 | 24,830 | 36,171 | 16,454 | | 10 | Feb 07, 2020 | 632 | 29,041 | 44,111 | 18,384 | | 11 | Feb 08, 2020 | 739 | 33,967 | 53,794 | 20,541 | | 12 | Feb 09, 2020 | 864 | 39,727 | 65,602 | 22,951 | | 13 | Feb 10, 2020 | 1,011 | 46,465 | 80,003 | 25,644 | | 14 | Feb 11, 2020 | 1,183 | 54,345 | 97,564 | 28,652 | | day | date | deaths | cases | @ lo dbl | @ hi dbl | |:--- | ------------:| ------------:| -------------:| -------------:| -------------:| | 1 | Jan 29, 2020 | 154 | 7,091 | 7,393 | 6,774 | | 8 | Feb 05, 2020 | 462 | 21,230 | 29,660 | 14,726 | | 15 | Feb 12, 2020 | 1,383 | 63,561 | 118,981 | 32,014 | | 22 | Feb 19, 2020 | 4,143 | 190,297 | 477,290 | 69,595 | | 29 | Feb 26, 2020 | 12,403 | 569,731 | 1,914,634 | 151,293 | | 36 | Mar 04, 2020 | 37,135 | 1,705,718 | 7,680,489 | 328,895 | | 43 | Mar 11, 2020 | 111,181 | 5,106,750 | 30,810,015 | 714,985 | | 50 | Mar 18, 2020 | 332,865 | 15,289,096 | 123,593,305 | 1,554,303 | | 57 | Mar 25, 2020 | 996,564 | 45,774,017 | 495,790,250 | 3,378,893 | | 64 | Apr 01, 2020 | 2,983,613 | 137,042,800 | 1,988,845,356 | 7,345,363 | | 71 | Apr 08, 2020 | 8,932,639 | 410,292,352 | 7,978,184,014 | 15,968,056 | | 78 | Apr 15, 2020 | 26,743,422 | 1,228,374,011 | 32,004,207,853 | 34,712,893 | | 85 | Apr 22, 2020 | 80,067,114 | 3,677,628,161 | 128,383,767,336 | 75,462,218 | | 92 | Apr 29, 2020 | 239,712,883 | 11,010,448,584 | 515,007,020,041 | 164,047,010 | | 99 | May 06, 2020 | 717,676,247 | 32,964,174,931 | 2,065,932,759,210 | 356,621,130 | Chart is formatted as a Markdown chart, BTW.

perl4ever 7y ago
"growth rate matters vastly more" Only in the case where you deny the existence of uncertainty in that number. Also, it's generally considered that infectious and lethal diseases rapidly evolve to become less virulent, since the people that are sickest tend to be more isolated.
- dredmorbius 7y ago
  No. The point is that initial quantity has virtualy no impact on final result, as compared with growth rate. With a doubling time of under two days, a misestimation by a factor of two ... is resolved in a day. By a factor of 10, in about 3 days. By a factor of 100, in six. By a factor of 1,000, in 12. It's not uncertainty about the growth rate, but change which matters, and which epidemiological containment is concerned with. The goal is to change the transmission rate (R factor), and through that, the rate of growth of the epidemic. Epidemics of the past which have proved highly lethal (look through your favourite list of historical plagues) have killed from 10% to half of local populations, often through a combination of high mortality but slow onset. Spreading before* symptoms are clearly expressed will suffice. The Black Death spread throughout Europe from 1346 - 1353, over eight years, killing up to 60% of the population. It returned multiple times, through 1667, with recurrence mortalities of 10-20%. It continued through the 19th century in the Arabic world. Expecting a novel virus to attenuate markedly in a few days or weeks seems wishful thinking.
  - perl4ever 7y ago
    "It's not uncertainty about the growth rate, but change which matters" You're saying growth rate is not rate of change and the rate of change is important, but the difference between two different rates of change is insignificant? My point was that the R factor is neither constant nor precisely known, whatever people do. If the outcome is very sensitive to the number, then it seems to me obvious that it is very important to look at the outcome over the range of possibilities, because it's going to be a large range. Whenever professionals make long term predictions, they usually depict them as a cone. How big it is and how it widens is important. Also, I thought the plague was bacterial, so I'm doubtful of the inference that a virus won't evolve rapidly. Previous flu epidemics did, right?