21
COVID-19 / Re: How do the Imperial modeling vs Oxford papers compare?
« on: 30/03/2020 17:00:39 »
The modelling paper from the Oxford group makes a large number of assumptions about the homogeneity of the population and the way the agent incubates and spreads. The result of their modelling is a prediction that as many as 50% of the population have already been infected.
This seems extremely unlikely. If the assumptions it makes were true, it's a mystery why America, extremely well connected as it is, is only just now joining the party; the virus would have spread there and delivered big numbers much sooner than it actually has. It's also a mystery how so many people (half the population) could be infected yet there be no corresponding spike in deaths like that which we are now seeing and what has been witnessed in Italy.
On the other hand, the Imperial College paper by Neil Ferguson and his colleagues models how a new agent, to which the population is not immune, is likely to spread. They make various predictions of case numbers over time, and they model 2 control scenarios: mitigation vs suppression (what we are doing now) and ask how they compare in terms of case load.
It's the Imperial study that has informed the government's present approach. Today, the same group have now released a new paper that uses rates of death to back-extrapolate to predict the likely number of cases in 11 countries across Europe. This suggests that 43 million infections have occurred across Europe, 1.7M of them in the UK. This means that the attack rate of infection (the number of people infected as a proportion of the whole) is about 3% in the UK, but much higher in Italy at close to 10%.
They then forward extrapolate to predict how many deaths have been averted by the present control manoeuvres. The answer is an estimated 59,000 people across Europe, 400 in the UK.
They acknowledge that serological assays (antibody tests) are urgently needed to clarify this situation and test the validity of these predictions.
This seems extremely unlikely. If the assumptions it makes were true, it's a mystery why America, extremely well connected as it is, is only just now joining the party; the virus would have spread there and delivered big numbers much sooner than it actually has. It's also a mystery how so many people (half the population) could be infected yet there be no corresponding spike in deaths like that which we are now seeing and what has been witnessed in Italy.
On the other hand, the Imperial College paper by Neil Ferguson and his colleagues models how a new agent, to which the population is not immune, is likely to spread. They make various predictions of case numbers over time, and they model 2 control scenarios: mitigation vs suppression (what we are doing now) and ask how they compare in terms of case load.
It's the Imperial study that has informed the government's present approach. Today, the same group have now released a new paper that uses rates of death to back-extrapolate to predict the likely number of cases in 11 countries across Europe. This suggests that 43 million infections have occurred across Europe, 1.7M of them in the UK. This means that the attack rate of infection (the number of people infected as a proportion of the whole) is about 3% in the UK, but much higher in Italy at close to 10%.
They then forward extrapolate to predict how many deaths have been averted by the present control manoeuvres. The answer is an estimated 59,000 people across Europe, 400 in the UK.
They acknowledge that serological assays (antibody tests) are urgently needed to clarify this situation and test the validity of these predictions.
The following users thanked this post: GoalTorrent