EpiMEC

Research

Six themes, one toolbox

Each theme below grew out of a public-health question in Costa Rica or in a partner institution. The tools travel between them: a bifurcation result from a relapse model informs a dengue model, and a Bayesian pipeline built for dengue was reused for COVID-19.

Vector-borne diseases

Dengue, Zika, chikungunya and malaria in a changing climate.

Costa Rica's dengue seasons have been the group's longest-running case study. We combine surveillance data with climate and environmental variables to build risk maps, forecast local incidence, estimate outbreak parameters and project dengue transmission to mid-century under climate scenarios.

Outbreak response

Projections and policy for COVID-19 in Costa Rica.

During the pandemic the group produced projections used in national decision-making: the effect of mobility and sanitary measures on community transmission, the impact of variants and vaccination strategies, multilayer network models of the country, and the role of testing on hospitalizations.

Behavior and social dynamics

Relapse, risk perception and behavioral epidemics.

Many health problems spread through social contact rather than pathogens. We model relapse and recidivism with nonlinear contact rates, analyze the forward-backward bifurcations that make such problems hard to eliminate, and study how localized risk perception changes an epidemic on a network. Current student projects model bulimia, adolescent pregnancy and population behavior.

Statistics and machine learning

From surveillance data to forecasts.

Bayesian spatio-temporal models with INLA, wavelet analysis, approximate Bayesian computation, generalized additive models and random forests: the statistical side of the group turns noisy epidemiological, climate and socioeconomic data into estimates with honest uncertainty.

Structured populations and networks

Age, patches, life stages and interconnected populations.

Real populations are not well mixed. We study epidemic models with age structure, short-term immigration, two-patch and source-sink dynamics, life stages, and stochastic dynamics across interconnected populations, and what these structures mean for thresholds and control.

Ecology and One Health

Disease in animal, plant and human systems.

The same tools apply beyond human epidemics: managing host populations after a disease-driven decline, biological control of coffee rust, and the interplay between foraging choices and population growth.

Methods

What is in the toolbox

  • Compartmental models · SIR-type systems with age structure, patches and life stages
  • Bifurcation analysis · forward-backward bifurcations, thresholds and control
  • Network models · multilayer and adaptive contact networks, risk perception
  • Stochastic dynamics · interconnected populations and extinction
  • Bayesian inference · spatio-temporal models with INLA, approximate Bayesian computation
  • Machine learning · random forests, generalized additive models, wavelets
  • Climate–disease coupling · high-resolution projections under climate scenarios
  • Numerical analysis · PDE models and general numerical implementations