If you like mathematics and statistics, the daily reports on COVID-19 and the generous supply of predictions, whether or not by experts, offers considerable food for thought. But even those who think of mathematics and statistics as very abstract disciplines, and who prefer to keep such disciplines at arm’s length, are actually in the middle of it. Each number corresponds to a son, a daughter, a partner, a granddad, a grandma, an aunt, a neighbor, a best friend,… who is infected by COVID-19 or recovered from it. In reverse, our behavior directly links to just how much chance the virus is given to spread among the population. All of us drive the figures! Our team provides the much-needed expertise around these figures.
Mathematical and statistical models help estimating important epidemiological quantities, such as the number of infected cases, the number of hospitalizations, the number of people in intensive care beds, and the number of deaths. Certain but not all models allow to cast short-term or even long-term predictions and to examine how such quantities change with changing human behavior and measures taken, such as the now well-known social distancing or vaccination programs.
Predictions are made according to a variety of scenarios, and are a great support for the team to offer well-founded advice to policy makers. Such advice is routinely based on more than one model. Each model has its strengths and pitfalls, and simultaneously considering various models strengthens prediction. Some models operate at macro level (e.g., to study the number of cases in the population of an entire country), while others operate at regional or local level. While some models pretty much rely on data only, others also use so-called model assumptions, often driven by the mathematical theory behind infectious diseases, knowledge from previous epidemics, etc. There is no such thing as the correct or best predictive model. Each model provides a piece of the jigsaw puzzle, and it requires a good amount of expertise and skill in infectious disease modeling to lay the entire puzzle.
Apart from modeling the epidemic, mathematics and statistics, i.e., numbers, are used in clinical trials conducted in an effort to find powerful antiviral medication, and ultimately vaccines.
What do we do?
Researchers (epidemiologists, statisticians, mathematicians, virologists, biologists, computer scientists, etc.) try to predict how the epidemic will evolve, and how this evolution changes under the influence of the measures that we take. The word “try” is here for a reason because the predictions produced by models are necessarily surrounded by a degree of uncertainty: models are never able to capture reality in all of its detail on the one hand, and also the data used are necessarily prone to imprecision and error. Data that we would like to have but that are lacking (such as the real number of new infections, rather than merely the number of confirmed cases) increase uncertainty. Certain key epidemiological parameters, such as incubation period, infectiousness as a function of age (children, adults, elderly), seasonality,… may be lacking at a certain point in time, or may be very imprecisely known. This is especially true in the beginning of an epidemic induced by a newcomer such as our SARS-CoV-2. The longer the time frame over which we want to make predictions, the more uncertain such predictions become (a bit like the weather forecast…).
The art and science of mathematical and statistical modeling is to translate the predictions obtained from a collection of models into advice useful for policy makers, who in term are the ones deciding on the measures taken, e.g., to “flatten the curve.”