While most of our time was spent on building the Mk I asset training system, Following the call from the Royal Society for modellers, we did lend some of our resource.
We used system dynamic models to simulate the impact of Covid in major cities (we were working with a global team investigating city level approached to mitigate Covid). Our responsibility was to model London and Stockholm.
One of the early problems was that the data was very poor and varied widely, so both the rate of reproduction Ro and the impact of the illness were not well known, so models could vary widely
It quickly became clear that some form of data sharing approach was needed and the NHS stepped up, using its social network (which we had worked on helping set up a few years earlier) to share data between analysts – and not just in the UK (this later became the Analyst X service)
London was typical of the problems faced with modelleing Covid – a lot of people (c 9 million), in close proximity to each other, with a wide variety of housing and occupancy types.
Ro itself comprises a number of parameters
- duration of contagion after a person becomes infected (how many people are infected at any one time)
- contact rate between people (how frequently does an opportunity to infect someone come up)
- likelihood of infection per contact with an infected person (estimating the % chance an uninfected person becomes infected in a contact)
In large, dense cities the contact rate is very high compared to rural or suburban areas, especially if there is a heavy reliance on public transport. It’s worse in colder countries as people spend more time indoors, and in expensive cities accomodation tends to be less spacious.
London was a high risk city. It was clear that stopping people contacting others was going to have the best impact and it was no surprise when the UK locked down. What was interesting is as data becem available the models showed that people had increasingly actually been “locking down” for at least a month before the official notification
Stockholm was the most intriguing as its approach was not to use any lockdown, an outlier in Europe. It’s pattern was different to London. (It’s a complicated story that needs another post)
At any rate, we completed our assignment, and were stood down. The UK authorities had chosen its favoured modelling teams (mostly the same ones they used beforehand…) but it was an interesting exercise and we hope we added some value.
The impact of the decisions made will probably not be clear for some years, it will be interesting to revisit London and Stockholm in a few years time