Realism

…for Malaria Program Analysts

While simple models are great for learning the basics, they are probably not up to the tasks of answering policy questions.

If we want build and analyze models to support malaria policies, we will need to have the capability of building models that are realistic enough to address policy questions. We need some way of setting rational expectations about the effects of malaria control and other factors that affect the health burden of malaria in a country, or in the health districts or villages. Some of the policy questions we must be able to address are:

These and other important questions about malaria can be addressed using some combination of mathematical models and data. Some of these questions would be difficult to address in any other way.

There are many other questions we will eventually need to address, but it is important to recognize that it is often much easier to ask a question than answer it. There are constraints on basic research that are unforgiving, so it would be challenging to design a study that could answer the questions that matter. If we are going to provide answers to these questions, we will need to use some combination of malaria research data collected over the past 144 years of studying malaria, the malaria surveillance data collected through national health management information systems (HMIS).

The analysis is challenging, and the answers will rarely by definitive.

Becoming an Analyst

It is a serious challenge to give good policy advice. Simulation-based analytics can help the analyst develop anlyses that can use information about the past to paint an accurate picture of malaria in the present and develop policies that will affect the future. A difference between research and analytics is that policy advice must be done on a schedule, and it is better to be informed by weak evidence than to act on no evidence at all. The rules for weighing uncertainty are different, and malaria analytics needs an inferential framework tailored to the needs of malaria programs. This is why we developed the notion of robust analytics for malaria policy, or RAMP. We envision using simulation-based analytics iteratively, with the goal of learning about malaria over time. These goals are described in another website, called Robust Analytics for Adaptive Malaria Control. Malaria theory provides important background for the malaria analyst.

We encourage anyone who is engaged in the process of building models for policy to think critically about the process of building models.

What makes a model good?

…and more importantly for malaria analysts:

What makes a policy analysis good?

There is a general sense that a good model is not too complex and not too simple – that it’s just right for the task at hand. It should be anchored to available data, and it should acknowledge various sources of uncertainty and find some way of dealing with that uncertainty.

In This Section

In these vignettes, we have developed a critique of simple models, and we go a bit further in our quest to understand, What makes a model good?

  • We present vignettes describing heterogeneity and complexity.

  • We believe it is important for malaria analysts to be able to critique various malaria models, so we have put an emphasis on developing some vocabulary for model-model comparison. In particular, we present idea of a skill set as a way of characterizing the scope and limitations of models.

  • Finally, we discuss why it makes sense to think about malaria as a complex adaptive system.

Up Next

The sections that follow cover five major areas:

  • Malaria Epidemiology in the narrow sense.

  • Transmission Dynamics by adult mosquitoes, including adult mosquito blood feeding behavior and demography, and parasite infection dynamics in mosquitoes.

  • Mosquito Ecology is a longer discussion of mosquito ecology, including exogenous forcing by weather, aquatic dynamics, and a longer discussions of adult mosquito behaviors.

  • Health Systems, including care seeking, surveillance,

  • Vector Control, including insecticide treated bed nets, indoor residual spraying, larval source management,

After that, we revisit the topic of Analysis, some Advanced Topics and a discussion of Models and Data.