Print this page
Published in News

UK computer modelling slammed

by on02 March 2021


Too complex

While computer modelling has driven much of the pandemic response, the way it is being used in the UK has been slammed.

Dr Thomas House [no really. Ed] , who sits on SPI-M - the group that models the pandemic for the government – told the Express that while models were an essential tool they were far from foolproof.

Despite the name, House is a mathematics reader at the University of Manchester and he is worried that people expect too much accuracy from models of a complex system.

“This does not mean we need to ditch models. You cannot do science without them. Modelling tells you about what might happen. Models can also be useful conceptual tools. However, the models are only as good as the assumptions that they are based upon. They may not account for many things such as seasonality, details of vaccine rollout individual behaviour.

“Nor can they account for our values and culture for example that people have different tolerances of risk. There are always values in politics and these need to be explained and it needs to be clear how the government has made their decisions.”

He said:  “It’s no good waving curves and graphs at people - if these only show COVID cases and deaths. We cannot be clear that the precautionary principle of first do no harm applies, because we don’t know the costs of lockdown measures.

Professor David Paton, an expert in health economics at Nottingham University said: “The use of models during this pandemic has been a great source of frustration. The government has relied too heavily on models predicting deaths, hospitalisations, and cases."

Professor Carl Heneghan, Director of the Centre for Evidence Based Medicine at Oxford University said: “At the moment we have a tug-of-war. Many argue the government has over relied on models which are then torn apart by critics. The key is not to be panicked by models. They are not a fait accompli.

"We need a new approach to integrate the emerging data with the models to keep them up-to-date so we can react appropriately as data emerges. We need to use models to help guide our thinking, but in a pandemic we also need to be flexible. The data is shifting all the time.

"In the last week deaths are down by a third. We should speed up our responses with the data as it emerges and this has not been happening. The world will look very different in two weeks' time and the idea that you can sit down and predict this or even predict that by mid-June we will be out of lockdown is not evidence based.”

Last modified on 02 March 2021
Rate this item
(0 votes)