Research › Computation

Simulating epidemics before they happen.

We translate biological and behavioural data into agent-based models, contact networks and Bayesian inference pipelines that let us test interventions before they're deployed in the field.

Methods

How we model transmission

The computational toolkit behind our epidemiological research.

Simulation

Agent-based modeling

Multiscale, agent-based frameworks for malaria and mpox developed through the Iso Lomso (STIAS) and Wiko fellowships, simulating individual-level transmission and pathogen mutation across networks.

Networks

Phylodynamics & contact networks

Combining sexual behaviour survey data with phylodynamics and agent-based models — the framework developed for HIV prevention research in our PhD dissertation at Stellenbosch University.

Inference

Bayesian inference & uncertainty quantification

Approximate Bayesian Computation and stochastic model calibration, methods developed during the AIMS MSc and applied to fit models to sparse, noisy surveillance data.

Optimization & Scientific AI

Optimal control & scientific machine learning

From dynamic-programming optimal control to today's AI4Mpox project (University of Toronto), building scientific AI to learn, predict and control complex disease systems.

Toolkit

Built for reproducible science

Our pipelines run on open, reproducible tooling — from statistical languages to purpose-built simulation and phylogenetics software — version-controlled and shared with our partner institutions for large-scale runs.

Languages & scientific stack

Python R MATLAB Bash NumPy SciPy Pandas

ML & Bayesian inference

scikit-learn TensorFlow PyTorch Stan R Shiny

Epidemic & ABM simulation

deSolve EpiModel NetLogo

Phylodynamics & genomics

BEAST RAxML MEGA Nextstrain

model := AgentBasedModel(pathogen="malaria")

network := ContactNetwork(nodes=12_400)

posterior := infer(model, data, method="MCMC")

> simulating interventions across scenarios …

> effective R estimate: 0.91

Curious about the models behind the numbers?

Reach out if you'd like to collaborate on modeling, share data, or discuss a joint project.

Get in touch