Read on Twitter
An interesting paper on lessons from Foot and Mouth outbreak in UK, and mathematical modelling:
https://www.research.ed.ac.uk/portal/en/publications/use-and-abuse-of-mathematical-models-an-illustration-from-the-2001-foot-and-mouth-disease-epidemic-in-the-united-kingdom(e368e2d3-1161-4e6e-97c8-5e33d44a9fb3).html
https://www.research.ed.ac.uk/portal/en/publications/use-and-abuse-of-mathematical-models-an-illustration-from-the-2001-foot-and-mouth-disease-epidemic-in-the-united-kingdom(e368e2d3-1161-4e6e-97c8-5e33d44a9fb3).html
"The UK experience provides a salutary warning of how models can be abused in the interests of scientific opportunism."
"Numbers, therefore, may convey an illusion of certainty
and security that is not warranted (43); for example,
because of the use of whatever numerical data are
available, regardless of their relevance and quality (38)."
and security that is not warranted (43); for example,
because of the use of whatever numerical data are
available, regardless of their relevance and quality (38)."
"The degree of confidence in the 2001 predictive
models is therefore low because they were not widely
tested,"
models is therefore low because they were not widely
tested,"
"The 2001 predictive models were constructed in an
environment of poor-quality data …, and poor
epidemiological knowledge (e.g. the transmission
characteristics of the virus strain, …). Therefore,
their use as predictive tools was inappropriate."
environment of poor-quality data …, and poor
epidemiological knowledge (e.g. the transmission
characteristics of the virus strain, …). Therefore,
their use as predictive tools was inappropriate."
"the models incorporated parameters that conflicted with the known biology of the virus"
"In retrospect, very little of value was added to the FMD
control policy by the use of predictive models."
"However, the consequences of following the
recommendations of these models were severe"
control policy by the use of predictive models."
"However, the consequences of following the
recommendations of these models were severe"
From Conclusions:
"Modelling should only be countenanced if … the design of the model and the information used to generate its results are correct (and plausible, from the known biology of the disease). Otherwise, models: ‘become exercises in
mathematical sophistry’"
"Modelling should only be countenanced if … the design of the model and the information used to generate its results are correct (and plausible, from the known biology of the disease). Otherwise, models: ‘become exercises in
mathematical sophistry’"
"the rift between the models and the practical reality of implementation may be so huge as to make the models irrelevant"
"The most appropriate use of models is as inter-epidemic tools, to aid retrospective analysis of real epidemics to gain an understanding of their behaviour."
I.e., as an educational and research tool, rather than a real-time policy tool.
I.e., as an educational and research tool, rather than a real-time policy tool.
"The use of models during epidemics should be restricted to
monitoring the epidemic and aiding short-term fine adjustments to strategies."
monitoring the epidemic and aiding short-term fine adjustments to strategies."
Ultimately:
"the utility of predictive models as tactical decision support tools is limited by the innate unpredictability of disease spread"
/fin
"the utility of predictive models as tactical decision support tools is limited by the innate unpredictability of disease spread"
/fin
