From modelling brain activity to modelling the impacts of COVID-19

Rosalyn Moran usually applies dynamic causal models to study brain networks, using data from imaging and other measurement techniques to explore what’s happening across populations of cells in our brains. The types of models she uses are very similar to those used in epidemiology and, as the COVID-19 pandemic broke, she decided to apply her skills to try to inform the discussion around the impacts of the virus on human populations.

‘The real trigger for me was talking to my sister who works in the Health Service Executive (Ireland’s NHS) in Dublin,’ says Dr Moran. ‘She asked if, with my background in modelling, there was anything I could do to help understand the current and future situation better. I started looking at the literature and different models that have been used in past and current epidemics and, when a colleague shared a model that they had developed, I ran it on the available data. That’s the beauty of this area: it’s very open and people often share the codes and the approaches they take.’

In collaboration with colleagues from King’s College London and UCL, she has published a study on Wellcome Open Research that is undergoing peer review and that estimates how deaths from the COVID-19 pandemic will fluctuate in coming months across seven European countries.

Following the introduction of measures to reduce the spread of COVID-19 such as social distancing, some countries are beginning to report a slowed rate of growth or a decline in the numbers of people testing positive each day for COVID-19. This suggests that countries are at or have past ‘the peak’ of infections. In order to provide insight into the future patterns or cycles of the pandemic, Dr Moran used a modelling approach to estimate the susceptible population, which is the number of people at the start of the outbreak who would eventually become infected.

‘When I first ran the code from various models on data from different countries I was interested to see how small the actual infected population was compared to the whole population,' Dr Moran said. 'I then started to wonder about estimating this susceptible population further down the line of the pandemic to see how this could help understand smaller peaks of infections that might happen in the future.'

Using different epidemiological models, including a new dynamic causal model developed by a former colleague, and assuming that those who have contracted COVID-19 and recovered from the disease are immune, the study estimated that on average across the countries 6.4% of the total population would be immune after the first peak. This percentage ranged from 19.6% of population in the UK to 4.7 % for Germany.

Based on these estimates and by re-running the models on the ever decreasing figures for susceptible populations after each peak, the study suggested that there would be several more peaks of infection in the future which get smaller and smaller until herd immunity has been reached. Herd immunity is when 60% of the population is immune and there is sufficient resistance to the spread of a contagious disease within a population. This is assuming that nobody has innate immunity, which too remains to be fully determined.

Image: Model predictions of cumulative deaths across countries from Wellcome Open Research

The study estimated that after about nearly 50,000 deaths from the first cycle in the UK there would be an additional 3 cycles before reaching herd immunity. In Germany the study estimated there would be only 5000 deaths in the first cycle but that they would experience the largest number of cycles at 12.

‘Due to the nature of our approach we can’t plot the timescale of the cycles in detail,’ comments Dr Moran. ‘But we can estimate how many cycles each country will experience and it seems that there is a bit of a trade-off between protecting the population which then leaves them susceptible to infection and the number of cycles or waves in the future, which some people call the ‘prevention paradox’. Maintaining current social distancing measures should prevent future waves and the cycles we modelled in this paper were based on the types of behaviours that triggered the first outbreaks back in January. A follow up study we have prepared that is a pre-print looks at waves in the context of social distancing which predict a less severe picture but show that the ‘new normal’ may be here for a while.' 

Dr Moran also highlights that now the government is carrying out antibody testing, the figures for susceptible population may be better known – which is crucial to improve the modelling. The study examined seven European countries: UK, Ireland, Germany, Spain, France, Italy and Switzerland and used data from the 2019 Novel Coronavirus at John Hopkins University Centre for Systems Science and Engineering on records from January 22^nd to April 5^th 2020.

Dr Moran and the team she is working with are now using the modelling approach to investigate the impact of other possible measures to contain the virus such as contact tracing alongside social distancing and also to explore modelling at a more regional level within the UK.

Image: Dr Rosalyn Moran. Credit: Virginia Tech Carilion School of Medicine and Research Institute

‘I think if you have something to offer in this situation and it’s not taking resources away from other areas then you should do what you can,’ Dr Moran comments. ‘In our line of work we have a computer infrastructure that we’ve built over 20 years and it can be really useful in handling this type of data. It’s also fantastic that we have these worldwide and regional datasets open so that everyone with relevant skills can analyse the data and bring a different perspective to the discussions.’

Reference: Moran et al. (2020) Estimating required ‘lockdown’ cycles before immunity to SARS-CoV-2: model-based analyses of susceptible population sizes, ‘S0’, in seven European countries, including the UK and Ireland Wellcome Open Research