It is well known that there is an interplay between the spread of infectious disease and the behavior of individuals in a population, and of course, the COVID-19 pandemic is not the exception. An outbreak can trigger behavioral responses, at both the group and individual levels, which in turn influences the epidemic evolution. These interactions can be addressed from an individual point of view as an adaptive co-evolutionary network1 or more explicitly through ordinary differential equation systems that incorporate the relevant parameters.
Always keeping in mind these feedback relationships between intervention and epidemic dynamics, we always try to predict what might happen with the course of the epidemic, although normally we do this retrospectively. Instead, in the current COVID-19 odyssey everything is made at real-time, therefore it is possible to find different difficulties such as the huge discrepancy in the way that data is reported, the great differences among countries, and of course the little we know so far (the level of generated immunity, which might appear not to be long-lasting is one good example).
Even with the aforementioned limitations and uncertainties, it was very important to propose models that could allow policymakers to figure out in the most realistic way, the possible evolution of the disease in the population. For example, how long should the activity (economic) lockout last, or how would the evolution of the outbreak affect these measures? So, our initial concern at the beginning of the pandemic was to figure out how long total population confinement should be extended in order to minimize the size of the epidemic peak2. Then we saw the necessity to explore other scenarios due to the rapid evolution of not only the pandemic but also the governments and population reaction to the virus propagation. This is when the complementary study begins, in order to cope out with the best way to ‘exit’ the lockdown situation. That is, trying to answer new questions such as evaluating the effectiveness of non-pharmacological interventions (use of masks and social distance), trying to produce actualized reports for policymakers in a very short time period because in these fast-moving diseases we know that time is money and more important, human lives to save.
We simulated different scenarios, as diverse as possible, as quickly as possible in a sort of frenetic race to try cover as much as possible, all range of possibilities that previous epidemics have left us as a lesson3. And of course, also trying to be conservative enough so as not to be accused of being an alarmist.
At the end of the day, the forecasts are shown to policymakers hoping that they will be used to improve the evolution of the disease in some way, hoping also to obtain some positive feedback from them, something that does not happen very frequently. But if this happens, and if forecasts are eventually taken into account and used to design effective control actions, one can ironically run the ‘risk’ of affecting positively the predicted evolution of the epidemic and of then being accused of having "failed" in its forecasting. But obviously this would be a minor side effect and one we should live with, as lucky us or not, they did not follow our advice!
All thanks to the aforementioned relationship between the spread of the virus and the population's (or sometimes political) behavior.
An special acknowledge to Xavier Rodó, the lead scientist of the Climate and Health program of ISGlobal.
Maria Cristina Avenue in Barcelona, empty. / FERRAN NADEU
Source: El periódico. Mar, 25 2020.
- López, L., Fernández, M., Gómez, A. & Giovanini, L. An influenza epidemic model with dynamic social networks of agents with individual behaviour. Ecol. Complex. 41, 100810 (2020).
- Lopez, L. R. & Rodo, X. A modified SEIR model to predict the COVID-19 outbreak in Spain and Italy: simulating control scenarios and multi-scale epidemics. medRxiv (2020) doi:10.1101/2020.03.27.20045005.
- Oxford, J. S. Origins of recent pandemics and lessons for surveillance. Influenza Surveillance 40–49 (2014) doi:10.2217/ebo.13.642.