Covid-19: why it is increasingly difficult to model the evolution of the epidemic – archyde

The latest results came out on October 8 computers from teams of modellers, which feed the public authorities on the future of the Covid-19 epidemic. With a “Cautious optimism”, they do not foresee a significant recovery before the beginning of the year 2022. But everything depends on the climate, the vaccination coverage, the effectiveness of the vaccines, the presence of variants, the relaxation of the French… A long list which testifies to the difficulty of grasping, for a year and a half, the epidemic upheavals and that sums up the warning which traditionally introduces these results: “These scenarios are made on the basis of incomplete data and uncertain assumptions. (…) The trajectories described depend on the assumptions made; if the assumptions do not materialize, the observed dynamics may be different from the projections. “

“We are often told that our models predict everything and its opposite and that they are useless, recognizes modeler Vittoria Colizza from the National Institute of Health and Medical Research (Inserm), whose work has informed public decision-making since the start of the pandemic. But they help us understand what might happen under different assumptions. “ With also the objective of “Make these models lie, so that the worst never happens”, emphasizes the researcher.

If the measures decided by the public authorities, in view of the projections, made it possible to avoid the worst scenarios, at several times during the pandemic, the models seemed disoriented because of many uncertainties. Or on the properties of the virus and its variants, whose transmissibility is difficult to assess in real time. Or on the behavior of the French themselves (respect for barrier gestures or relaxation, risky behavior or over-precaution, etc.), which are difficult to model and anticipate, whereas they are central to estimating the speed of the epidemic’s spread . Or, again, on political decisions, which do not always fall at the time chosen by the modelers for their calculations.

January-February 2021: a turning point difficult to capture

the 29 january, the modellers from the Institut Pasteur and Inserm publish a note to describe the possible scenarios in France at the end of winter, taking into account the presence of last variant arrived, says Alpha. The current growth is expected to continue and only a new containment would be able to control the epidemic. But patatras, a few days later, the epidemic is on the contrary in rapid decline. “Our models were calibrated on hospitalizations, but the impact of the curfew was not yet visible”, emphasizes Vittoria Colizza.

You have 76.8% of this article left to read. The rest is for subscribers only.