Skip to contents

RAMP and SimBA

Robust, Simulation-Based Analytics for Malaria Policy

Source: vignettes/RAMP.Rmd
RAMP.Rmd

Robust Analytics for Malaria Policy (RAMP) was developed as an inferential system bespoke for malaria policy. Managing malaria is challenging because it is a set of loosely coupled, locally peculiar, complex adaptive systems. Mathematical models are a useful way of understanding the non-linear interactions among humans, mosquitoes, parasites, and malaria managers that make up these malaria systems. The goal of Simulation-Based Analytics (SimBA) is to use dynamical systems describing malaria in populations well enough to develop policy advice to help managers make effective policies. The goal of RAMP is to develop advice that is robust to uncertainty — in practical terms, this involves going to great lengths to characterize and quantify uncertainty and then to propagate the uncertainty through analytics pipelines.

The stem name ramp was adopted for a suite of R packages:

  • SimBA is a suite of packages designed to support robust, simulation-based analytics. Core computation is handled in ramp.xds, but it is supported by six satellite packages.

  • Two packages that take deep dives.

ramp.func — a function library — supports SimBA and both deep dive packages.

A diagram of the R packages and their dependencies. Clicking on the package icon takes you to its github.io page. SimBA is also a website (grey) with material to support model building.

SimBA

To support development of professional analytics in malaria-endemic countries, we developed a new mathematical framework that would facilitate nimble model building by analysts embedded in malaria programs in endemic countries.

To be nimble:

  • The software would need to be capable of building models with arbitrary levels of realism and complexity, a feature called scalable complexity.

  • It should be possible for a malaria analyst to solve most models in a reasonable amount of time using their work computer.

  • It should be straightforward to modify a model — to make it simpler or more elaborate — or to build a new model, as needed, to suit the changing needs of a malaria program.

  • All analysis should be transparent and replicable.

The software implements a mathematical framework that was developed in two phases. In the first phase, we developed a mathematical framework for modular, spatial dynamics with a focus on autonomous systems. In the second phase, we extended the framework to handle non-autonomous systems with exogenous forcing by weather, vector control, and mass health interventions.

SimBA is the set of seven R packages

  • ramp.xds handles all core computation, but its utility is extended by several satellite packages:

  • ramp.library is a dynamic module library. Each module implements a dynamical system describing the epidemiology of malaria or another mosquito-transmitted pathogen, adult mosquito ecology and infection dynamics, or aquatic mosquito ecology. It includes a large set of model families published in the peer-reviewed literature that are not included in ramp.xds. The ability to reuse code reduces the costs of replicating studies. Through this library, ramp.xds also supports nimble model building and analytics for other mosquito-borne pathogens.

  • ramp.func is a function library for RAMP.

  • ramp.forcing is a collection of utilities to model exogenous forcing by weather, vector control, and or mass health interventions.

  • ramp.demog handles human demography and stratification, including vital dynamics and age structure.

  • ramp.work includes algorithms to apply the framework, including code to fit models to data.

  • ramp.qa has method for qualitative analysis, including thresholds, scaling relationships, and connectivity.

Also See

  • Wu SL, Henry JM, Citron DT, Mbabazi Ssebuliba D, Nakakawa Nsumba J, Sánchez C. HM, et al. (2023) Spatial dynamics of malaria transmission. PLoS Comput Biol 19(6): e1010684. https://doi.org/10.1371/journal.pcbi.1010684.

Deep Dives

Two packages take deep dives into malaria epidemiology and mosquito ecology.

ramp.falciparum — Falciparm Malaria Epidemiology

Falciparum malaria epidemiology is extremely complex, even without considering transmission. Models must handle exposure, the complex time course of an infection (including parasite densities), superinfection, infectiousness, disease, treatment and chemoprotection with antimalarial drugs, immunity, and diagnostics and detection. This is a sandbox for falciparum malaria epidemiology with a mix of old and new models, including a probabilistic approach to falciparum malaria epidemiology, several queueing models describing malaria superinfection dynamics, development of hybrid variables and enhanced compartmental models, and some new models.

Also See

ramp.micro — Microsimulation

While ramp.xds was developed to handle patchy spatial (i.e. metapopulations), we developed ramp.micro to explore models for mosquito ecology and pathogen transmission dynamics on point sets.

Also See