On 14 March 2019, the RSS Merseyside local group hosted an event based on perspectives on risk with speakers coming from a range of different fields. The meeting had around 32 attendees, approximately half of whom were RSS members. The event attracted people from a range of institutions including Pathology at Royal Liverpool University Hospital, Seqirus, and Institute of Risk and Uncertainty and the Departments of Biostatistics, Mathematical Sciences, Molecular and Clinical Medicine from the University of Liverpool.
The meeting began with a talk by Dr Jet G Sanders from the London School of Economics who discussed how our risk tolerance fluctuates with the weekly cycle. This was a fun talk and risk aversion was illustrated by getting attendees to compete to see who could blow up the biggest balloon with the risk of it popping increasing with the size of the balloon.
Jet went on to illustrate how a pattern of risk tolerance decreasing from Monday to Thursday and then having a significant increase on Friday is seen across a wide range of events, from blowing up a balloon to robbing a bank. By understanding the weekly fluctuations in risky behaviour at a population level, there is the potential for improvements in our national systems. For example, Jet demonstrated that voting intentions may change according to the day of the week, and that including the day of the poll as a variable can improve predictions of election or referendum outcomes.
This was followed by a talk by Dr Edoardo Patelli from the Risk Institute, University of Liverpool, who spoke about risk in engineering and the problem of missing data or a lack of data. In this case, standard assumptions are often made, however he highlighted that the common assumption of a uniform distribution across an interval, often thought to be uninformative, is a large assumption which can lead to inaccurate measures of confidence. Generalised probabilistic methods are a way to avoid making any assumptions when dealing with sparse data, and allow for different representations of uncertainty. These methods were particularly successful in addressing the NASA Langley Generic Transport Model challenge which aimed to provide estimates of uncertainty for new systems and vehicles operating in extreme conditions.
Finally, Dr Helen Clough from the University of Liverpool discussed how norovirus is a huge under-reported risk with the true burden of infection not well recorded. Three different public bodies collect data on norovirus outbreaks associated with seafood, and the overlap between these three data sets is very varied. Capture-recapture methods, more commonly associated with ecology, can be used here to provide estimates of the underlying rate of norovirus infection from seafood. These methods allow for data from more than one organisation to be included by accounting for relationships between organisations data collection methods, and therefore provide greatly improved estimates of disease burden. This highlights the need for different organisations to integrate their approaches to surveillance.
The next meeting of the RSS Merseyside local group will take place on Monday 13 May on the topic of Mental Health Statistics. Confirmed speakers are Dr Anais Rouanet (MRC Biostatistics Unit, University of Cambridge) and Professor Richard Emsley (Kings College London). Further details of this, and all future talks, are available on the Merseyside local group website: https://sites.google.com/site/rssmerseyside/