However, given this announcement I thought it would be interesting to look where not to visit in the coming weeks.  Below are the top ten ‘flu locations world-wide — based on how likely are you to bump into someone who is infected.

1. Chile   (11.6 cases per million per day over the past week)
2. Austrialia (4.7 cases per million per day over the past week)
3. Canada (3.9 cases per million per day over the past week)
4. Panama (2.7 cases per million per day over the past week)
5. Cayman Islands (2.6 cases per million per day over the past week)
6. Costa Rica (1.7 cases per million per day over the past week)
7. USA (1.5 cases per million per day over the past week)
8. Dominican Republic (1.3 cases per million per day over the past week)
9. Iceland (1.1 cases per million per day over the past week)
10. Dominica (1.1 cases per million per day over the past week)

Some interesting points to note. Firstly, Mexico is not on the list, its 5th in terms of total cases per person to date, but for now the epidemic seems to be decreasing. Secondly, the top two countries are both in the Southern Hemisphere and therefore in the middle of their winter flu season — this is what we predicted over a month ago (see Blog from 2nd May). The vast majority of the others are close to the orginal source in Mexico. The one big exception in Iceland — which is very surprising.

So should we be worried? Not really. Even if you met 25 new people per day while you were on holiday for a fortnight in Chile, then your chance of contracting the infection is still less than 1:100

Cases in the UK as of 13 May 2009

It is important to note that there is a lag between the time of infection and case confirmation. In the next few days this picture is likely to change, as there are over 300 suspected cases being investigated by the HPA in the laboratory.

Here we use two graphs to highlight the differences between different regions. The world-wide cases, fuelled by the epidemics in USA and Mexico have a steep exponential rise, with cases doubling every 5-6 days (much slower than 10th May when the doubling time was 2-3 days). In contrast, Canada, Europe and the UK show a much slower rate of increase, doubling every 9, 14 and 5 days respectively (comparable to the values from 10th May).

So does this mean that the influenza outbreak is no-longer a threat? Well, quite simply it is too early to say. The good news is that the doubling time for the number of cases has reduced dramatically in the USA; the bad news is that the number of cases is still increasing.

We now perform our 3-day prediction, based on a simple exponential fit to the recent data on laboratory confirmed cases. We also consider how well our prediction (made from data available on 11th May) performed. We now provide some measure of confidence interval based upon multiple exponential fits using different time lags and different weightings.

One thing we are very pleased to see is that the true cases in the USAare much lower than we predicted — there has been a clear change in the growth of infection here. However cases continue to rise, and our predictions for the UK and Europe were surprisingly accurate.

Here we use two graphs to highlight the differences between different regions. The world-wide cases, fuelled by the epidemics in USA and Mexico have a steep exponential rise, with cases doubling every 2-3 days. In contrast, Canada, Europe and the UK show a much slower rate of increase, doubling every 7, 13 and 5 days respectively.

One region that will be informative to watch over the coming days is Australia. Australia is currently in their winter (and therefore ‘flu season). Two main questions emerge: will the epidemic increase rapidly in Australia? will the new H1N1 cases be masked by cases of seasonal influenza?

We now perform our 3-day prediction, based on a simple exponential fit to the recent data on laboratory confirmed cases. We also consider how well our prediction (made from data available on 7th May) performed.

We therefore see that our predictions for the world and the world excluding Mexico were lower than the actual value, while we over-estimated the cases in Canada and Europe. However, our predictions for the UK and USA were surprisingly accurate.

The data is taken from the extremely helpful google-document compiled by the Guardian newspaper. We note that different sources provide slightly different views on the current magnitude of the epidemic. For example, looking at the total confirmed cases:

Using this simple mathematical model, we can predict the total number of confirmed cases into the future:

These predicted values make several major assumptions: most notably that the epidemic progresses unchecked for the next three days.

A few days ago we projected that  a strongly controlled outbreak would produce around 100 cases and last around 30 days, with a 25 % probability of developing into a country wide epidemic. These projections were made using the inital conditions of the model corresponding to the known cases up until the 29th of April. If we now introduce further known cases, we are more likely to see that the  infection run for longer, with around 50% of the simulations continuing for longer than 100 days, and causing more that 500 cases.

These projections are made under the assumption that no further cases will be introduced from outside the UK, which is an unlikely scenario, particularly as the US and the rest of Europe begins to have more and more cases. As more information becomes available, we will be updating this page and our projections.