Constructivism

Theory for Malaria Analysts

In developing a website to explain theory to malaria analysts, we adhere to strict constructivism. What is a constructivist? A branch of mathematical philosophy, called constructivism, takes a strong position about the need to have an example of a mathematical object to prove it exists. When we call ourselves constructivists, we mean that everything we explain here can be computed in some form and incorporated into an analysis. We will be using SimBA software suite, which was developed for these tasks.

Our constructivist ethos goes a bit deeper. In taking on the job of explaining malaria theory to malaria analysts, we are explicitly rejecting some academic baggage, and encouraging the malaria analyst to adopt a different ethos. Why? We are huge fans of scholarship, but many scholars pursue ideas for the pure joy of it, not to develop something useful. Malaria theory can be beautiful and elegant, and scholarship in malaria theory is needed, but this site was developed for analysts. Malaria is a disease that kills hundreds of thousands of humans each year. We care deeply about the task of reducing the burden of malaria and eradicating Plasmodium falciparum from the planet. In the service of that cause, we have taken a critical look at the whole business of academics doing global health, and it seems the drive to get grants funded and to publish papers in peer review has left some serious blind spots. In the service of giving robust policy advice, it is our task to build software that can deal with some of those blind spots. Not surprisingly, the blind spots surround some of the messiest aspects of malaria.

In taking the broad view of malaria theory, we are guided by some ideas from literary theory from Simulacra and Simulation, by Jean Baudrillard. The book discusses the concept of simulacra, or copies that depict things that either had no original, or that no longer have an original. Baudrillard discusses the process by which symbols and signs have come to replace human experience. In Baudrillard’s critique, life has become a simulation. Baudrillard’s ideas have had their greatest effect in philosophy and the arts, but we can examine science with the same critical ideas. The best way to avoid creating simulacra is to find ways of grounding the study in reality.

In the case of malaria analytics, it is critical to ensure that our simulations depict reality, but we are giving advice about a complex system, and we are often lacking most of the data we need to describe the processes. If an analyst offers advice that has not explicitly considered the messy process, then the analysis has probably made some implicit assumptions about it that they have not explained. While there is some value to writing academic papers in this mode, there is a difference between becoming an analyst and doing analysis to inform policy as a scientist.

There is a difference between the goals and activities to:

The former is an abstract description of a process. The latter – a model of malaria on Bioko Island, Equatorial Guinea, for example – is about a particular thing.

In doing the first thing – building a model that would apply anywhere – it is important to emphasize general principles. These sorts of models are published in peer reviewed articles, and they tend to be simple, elegant and abstract. These models have supported development of many academic careers, and malaria theory is drawn mainly from these sorts of models.

In doing the second – building models to manage malaria – it is important to develop an understanding of malaria in some local context to identify the relevant details affecting malaria epidemiology, transmission dynamics, and control here and now. These models are developed for decision support, and they will often need to be much more realistic and complex. The simple and elegant models that comprise malaria theory must be combined in some ways to resemble malaria in a place, and the resulting analysis might be relevant only in one place.

Both approaches are needed, and there are some potential synergies. On the one hand, theory can help structure model building for malaria analytics to give sharp insights into malaria transmission in situ. On the other, models developed for malaria managers can draw attention to critical gaps in malaria theory.

When we learn to build models of malaria, we start simple. We are learning principles, and it doesn’t matter how realistic they look. We’re trying to get to the gist of the matter. The cost of sticking with this approach is that that we an only ever give generic advice. At some point, we need to go beyond the simple models and add enough realism to get it right enough to manage malaria. We need a way of adding in realism.

Until recently, if you wanted to build a model that was realistic enough to simulate malaria, you would use a comprehensive individual-based malaria model (IBM). The complexity of malaria, the need to have models for integrated malaria contol, and funding from the Gates Foundation spurred development of several comprehensive individual-based models to answer questions through simulation. For the past decade, it has been possible to simulate malaria in situ with virtually unbounded realism using IBMs. An alternative system now exists, called SimBA (short for Simulation-Based Analytics).

If a model represents malaria epidemiology, transmission dynamics, and control accurately, then the analysis ought to point us in the right direction. Since the amount of data available to calibrate realistic models is limited, it’s hard to know if we have got it right. Drawing from the study of mimetics, we note the parallels to hyperrealism. Using AI, it is now trivial to create images that look like photographs, but the thing never existed. Some artists, poking fun at hyperrealism, have created highly realistic looking portraits of uncanny and grotesque monstrosities. The question applies to malaria, as well. How do we know if a model is an uncanny monstrosity, or if it is close enough to reality to put trust in the results? Without some theory, it is difficult to critique analysis of any complex model.

Theory is developed to help us understand how models behave and the limitations of those models. Without some understanding of the theory, malaria analyst are limited to reporting on the outputs of a simulation study. With a sound understanding of theory, the analyst has greater capacity to critically evaluate the model, and perhaps help malaria programs modify their surveillance to test working hypotheses or fill in critical data gaps.