just for laugh, here is the US H1N1 confirmed case count by date as provided by the CDC
now with added bonus WHO worldwide case count.
click to embiggen
looks pretty linear now, I think the dip at day 7 is a CDC error and the case number was 109 then, not 91.
Doubling time is roughly 2 days, or e-folding of about 3 days.
The early rise is catching up on existing cases.
There is some sign of flattening, could be rate slowing, could be CDC lab falling behind.
If the doubling time is really about 2 days, and the confirmed case count is not dominated by the ability of the labs to keep count, then the cases will increase roughly tenfold per week.
So thousand cases end of this week; million cases by the end of May.
That is a very big "IF", extrapolating from a very short data set with large selection biases.
UPDATE: here is the WHO confirmed case count worldwide
click to embiggen
from WHO situate update
NB: WHO is updating twice per day right now.
Also consistent with approximately 2 day doubling time...
- Log in to post comments
More like this
Dear Steinn,
Here is my fit of the WHO data: 841*(1.51)^t, where t is given in days (zero is today, -1 is yesterday etc). Of course this scaling is irrelevant for exponentials. The plot is here:
http://comciencias.blogspot.com/2009/05/gripe-suina-o-numero-de-casos.h…
The point is: the branching factor (at least for the WHO data) is around 1.5. The true branching factor, including the mild cases, is above this?
that is an exponential - any fit of the form n^t is equivalent to an exponential
looks like your fit also give a doubling time of about 2 days
Starting from the first WHO report (april 24)as t=0, where t is counted in days, we get the simpler fit:
N(t) = 20.35 * 1.52^t = 20.35*exp(ln 1.52 * t) = 20.35*exp(0.42 t).
The doubling time is given by 1.52^T = 2, that is, T = ln(2)/ln(1.52) = 1.7 days
This means 100.000 cases in May 15 and 100 millions in 31 May, as you have noticed. Of course, at this time the sigmoidal saturation will appear. Of course, the curve will fit the data only during the time that laboratories can cope with the identification task of new cases.
yup, that looks about right
the flattening to apparent linearity of the US cases is likely due to the CDC labs being maxed out and only doing fixed number of confirmations per day
if that is the case there should be a jump in known cases this week as the new primers go out to the certified regional labs and the throughput goes up, specially given the known backlog
Taking inspiration from this post, I made a more current graph of the WHO infections/deaths data: http://mightymu.net/doom.html
Updating daily from WHO.