Skip to content


Mathematical epidemiology, the COVID-19 pandemic and the limits and uncertainties of models

I have been in isolation for the past week or so, because my immune system has always been compromised. I am doing this against the current wisdom provided by the Mexican secretariat of health, but following global approaches to facing the COVID-19 pandemic, because apparently, “these kinds of measures are not necessary at the moment” (ask me later why I am so pissed off at Mexico’s epidemiological strategy against COVID-19). Anyhow…

In this blog post, I want to address some of the limitations of models and modeling, and the risk that we are currently facing of everyone wanting to become “arm-chair epidemiologists” in light of COVID 19. I write this as someone with a double PhD in political science and human geography who has spent years of his life reading, learning and understanding mathematical epidemiology models, and those three degree theses (undergrad in chemical engineering, Masters in economics and PhD) are models of various kinds (2 mathematical, 1 game-theoretic).

I am not an epidemiologist, nor do I have a PhD in epidemiology, and when I mention things like rate of transmission, mortality, etc. I do so establishing clear boundaries on what I think I CAN know. I’ve mentioned before how when I came in contact with mathematical epidemiology.

I started reading up on mathematical epidemiology around the early 2000s (20 years, now), and right after we got hit with several epidemics, including SARS, H1N1, MERS and Ebola. I sought experts on Twitter on this kind of modeling, such as Dr. Maia Majumder and Dr. Sherri Rose.

I understand well enough (thanks to having done mathematical modeling myself) that models, and particularly epidemiological models, rely on assumptions. What makes mathematical epidemiology important, in my view, is the combination of knowing math AND epidemiology. This area is inherently cross- and inter-disciplinary. Mathematical epidemiologists are modelers who understand the epidemiological assumptions and consequences of their models. Many of them, alongside virologists, statisticians, physicians, social workers, medical anthropologists, are losing sleep trying to figure out SARS-CoV2 (COVID-19). Modelers are trying to get the model assumptions right because they know that there’s enormous uncertainty in creating models. We calibrate models with empirical data for this reason, trying to adjust them to reality.

BUT, and here’s the but: reality right now is extraordinarily uncertain, and that’s what is making mathematical epidemiologists (and epidemiologists overall) so concerned. The natural sciences’ component of this epidemics are still developing. We can’t say “this is another flu”. Because there’s so much uncertainty (not only in the models we are using but also on the virology/epidemiology of this pandemics), we need to take precautions that, to some people, may look extreme. People are preaching social distancing and pre-isolation in abundance of caution. This is NOT just another flu and virologists/epidemiologists are trying to figure out why. We need interdisciplinary work where we acknowledge our knowledge limitations, the assumptions we are making, the potential risk pathways that we may need to walk, trade offs we make.

here’s a reason why public understanding of science scholars, communications specialists, risk analysis and disaster management academics are all face-palming right now (myself included). Not everyone is, nor should be an epidemiologist. We need specialists of all areas. We need to develop wiser ways to communicate the risks of COVID and the implications of models we make, and as consumers of this information, we need to accept that models bring along uncertainties, and our activities carry a certain degree of risk. This is all about managing it.

When people say “we are all in this together”, it’s because we are. COVID-19 is revealing why interdisciplinary work matters and our planet is becoming a global laboratory for its implementation. We all, civilians and scientists, have a role and responsibility in surviving this. This is the time to absorb different perspectives on the epidemiology and virology of COVID-19, and the human dimensions of this disease and the potentially negative implications it will have not only on financial markets but also on hospitals and local health systems. It’s vital that we understand what happens within such a complex system.

Models have assumptions and limited predictive power under contexts of high uncertainty. If you need reassurance as this issue develop, follow knowledgeable people and ask about the limitations, assumptions, and implications of a model. Trust me when I say that nobody is more oncerned about getting this right than the natural and social scientists (as well as the people in the humanities and math, communications specialists, science communicators, journalists). But it has to be an inter-, and cross-disciplinary.

Not just models.

You can share this blog post on the following social networks by clicking on their icon.

Posted in academia, research, research methods.

Tagged with , , .


One Response

Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.

  1. Rima says

    Interesting post, professor Pacheco.
    There is a big gap between the disciplines you re naming. The cooperation brtween these disciplines is rarely adopted, because researchers are often very suspicious and are most of the time willing to get the credit for the performed work. That is very human indeed. This is a period of chaos, and researchers do not have yet necessary hindsight for being able to come up with reliable models.



Some HTML is OK

or, reply to this post via trackback.



shares